repo
stringlengths
2
99
file
stringlengths
13
225
code
stringlengths
0
18.3M
file_length
int64
0
18.3M
avg_line_length
float64
0
1.36M
max_line_length
int64
0
4.26M
extension_type
stringclasses
1 value
tvm
tvm-main/python/tvm/meta_schedule/testing/tune_onnx.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import json import logging from distutils.util import strtobool import onnx # type: ignore import tvm from tvm import meta_schedule as ms from tvm.meta_schedule.testing.custom_builder_runner import run_module_via_rpc from tvm.relay.frontend import from_onnx from tvm.support import describe from .tune_utils import create_timer, generate_input_data def _parse_args(): args = argparse.ArgumentParser() args.add_argument( "--model-name", type=str, required=True, ) args.add_argument( "--onnx-path", type=str, required=True, ) args.add_argument( "--input-shape", type=str, required=True, help='example: `[{"name": "input1", "dtype": "int64", "shape": [1, 1, 8]}]', ) args.add_argument( "--target", type=str, required=True, ) args.add_argument( "--num-trials", type=int, required=True, ) args.add_argument( "--rpc-host", type=str, required=True, ) args.add_argument( "--rpc-port", type=int, required=True, ) args.add_argument( "--rpc-key", type=str, required=True, ) args.add_argument( "--work-dir", type=str, required=True, ) args.add_argument( "--number", type=int, default=3, ) args.add_argument( "--repeat", type=int, default=1, ) args.add_argument( "--min-repeat-ms", type=int, default=100, ) args.add_argument( "--adaptive-training", type=lambda x: bool(strtobool(x)), help="example: True / False", default=True, ) args.add_argument( "--cpu-flush", type=lambda x: bool(strtobool(x)), help="example: True / False", required=True, ) args.add_argument( "--backend", type=str, choices=["graph", "vm"], help="example: graph / vm", required=True, ) parsed = args.parse_args() parsed.target = tvm.target.Target(parsed.target) parsed.input_shape = json.loads(parsed.input_shape) parsed.rpc_config = ms.runner.RPCConfig( tracker_host=parsed.rpc_host, tracker_port=parsed.rpc_port, tracker_key=parsed.rpc_key, session_timeout_sec=600, ) return parsed logging.basicConfig( format="%(asctime)s.%(msecs)03d %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S" ) logging.getLogger("tvm.meta_schedule").setLevel(logging.DEBUG) ARGS = _parse_args() def main(): describe() print(f"Workload: {ARGS.model_name}") onnx_model = onnx.load(ARGS.onnx_path) shape_dict = {} for item in ARGS.input_shape: print(f" input_name : {item['name']}") print(f" input_shape: {item['shape']}") print(f" input_dtype: {item['dtype']}") shape_dict[item["name"]] = item["shape"] mod, params = from_onnx(onnx_model, shape_dict, freeze_params=True) input_data = { item["name"]: generate_input_data(item["shape"], item["dtype"]) for item in ARGS.input_shape } with ms.Profiler() as profiler: database = ms.relay_integration.tune_relay( mod=mod, target=ARGS.target, params=params, work_dir=ARGS.work_dir, max_trials_global=ARGS.num_trials, num_trials_per_iter=64, runner=ms.runner.RPCRunner( # type: ignore rpc_config=ARGS.rpc_config, evaluator_config=ms.runner.EvaluatorConfig( number=ARGS.number, repeat=ARGS.repeat, min_repeat_ms=ARGS.min_repeat_ms, enable_cpu_cache_flush=ARGS.cpu_flush, ), alloc_repeat=1, ), cost_model=ms.cost_model.XGBModel( # type: ignore extractor=ms.feature_extractor.PerStoreFeature(), adaptive_training=ARGS.adaptive_training, ), strategy=ms.search_strategy.EvolutionarySearch(), ) lib = ms.relay_integration.compile_relay( database=database, mod=mod, target=ARGS.target, params=params, backend=ARGS.backend, ) print("Tuning Time:") print(profiler.table()) run_module_via_rpc( rpc_config=ARGS.rpc_config, lib=lib, dev_type=ARGS.target.kind.name, args=input_data, continuation=create_timer(ARGS.backend), backend=ARGS.backend, ) if __name__ == "__main__": main()
5,537
27.111675
100
py
tvm
tvm-main/python/tvm/meta_schedule/testing/custom_builder_runner.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Customized builder and runner methods""" # pylint: disable=import-outside-toplevel from typing import TYPE_CHECKING, Dict, List, Optional, Union, Callable if TYPE_CHECKING: import numpy as np # type: ignore from tvm.ir import IRModule from tvm.meta_schedule.runner import EvaluatorConfig, RPCConfig from tvm.runtime import Device, Module, NDArray from tvm.target import Target from tvm.runtime.vm import Executable def build_relay( mod: "IRModule", target: "Target", params: Dict[str, "NDArray"], ) -> "Module": """Build a Relay IRModule Parameters ---------- mod : IRModule The Relay IRModule to build. target : Target The target to build the module for. params : Dict[str, NDArray] The parameter dict to build the module with. Returns ------- mod : runtime.Module The built module. """ from tvm.relay.build_module import _build_module_no_factory as relay_build from tvm.runtime import Module result = relay_build(mod, target=target, target_host=None, params=params) assert isinstance(result, Module) return result def build_relay_with_tensorrt( mod: "IRModule", target: "Target", params: Dict[str, "NDArray"], ) -> "Module": """Build a Relay IRModule with TensorRT BYOC Parameters ---------- mod : IRModule The Relay IRModule to build. target : Target The target to build the module for. params : Dict[str, NDArray] The parameter dict to build the module with. Returns ------- mod : runtime.Module The built module. """ from tvm.ir.transform import PassContext from tvm.relay.build_module import _build_module_no_factory as relay_build from tvm.relay.op.contrib import tensorrt from tvm.runtime import Module mod = tensorrt.partition_for_tensorrt(mod, params) with PassContext(opt_level=3): result = relay_build(mod, target=target, target_host=None, params=params) assert isinstance(result, Module) return result def run_with_graph_executor( rt_mod: "Module", device: "Device", evaluator_config: "EvaluatorConfig", repeated_args: List["NDArray"], ) -> List[float]: """Run a Relay module with GraphExecutor Parameters ---------- rt_mod : Module The Relay module to run. device : Device The device to run the module on. evaluator_config : EvaluatorConfig The evaluator configuration to run the module with. repeated_args : List[NDArray] The list of repeated arguments to run the module with. Returns ------- results : List[float] The list of results. """ import itertools from tvm.contrib.graph_executor import GraphModule graph_mod = GraphModule(rt_mod["default"](device)) evaluator = graph_mod.module.time_evaluator( func_name="run", dev=device, number=evaluator_config.number, repeat=evaluator_config.repeat, min_repeat_ms=evaluator_config.min_repeat_ms, f_preproc="cache_flush_cpu_non_first_arg" if evaluator_config.enable_cpu_cache_flush else "", ) repeated_costs = [] for args in repeated_args: profile_result = evaluator(*args) repeated_costs.append(profile_result.results) costs = [float(cost) for cost in itertools.chain.from_iterable(repeated_costs)] return costs def run_module_via_rpc( rpc_config: "RPCConfig", lib: Union["Module", "Executable"], dev_type: str, args: Union[Dict[int, "np.ndarray"], Dict[str, "np.ndarray"]], continuation: Callable, backend: Optional[str] = "graph", ): """Execute a tvm.runtime.Module on RPC remote""" # pylint: disable=import-outside-toplevel import os import tempfile from tvm.contrib.tar import tar from tvm.runtime import ndarray # pylint: enable=import-outside-toplevel with tempfile.TemporaryDirectory() as tmp_dir: filename = os.path.join(tmp_dir, "tvm_tmp_mod." + tar.output_format) if backend == "vm": code, lib = lib.save() lib.export_library(filename, tar) session = rpc_config.connect_server() session.upload(filename) _, filename = os.path.split(filename) rt_mod = session.load_module(filename) if backend == "vm": rt_mod = session.get_function("runtime.Load_Executable")(code, rt_mod) dev = session.device(dev_type=dev_type, dev_id=0) nd_args = {k: ndarray.array(v, dev) for k, v in args.items()} return continuation(rt_mod, dev, nd_args)
5,448
30.316092
83
py
tvm
tvm-main/python/tvm/meta_schedule/testing/dummy_object.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Dummy objects for testing.""" import random from typing import List, Optional from tvm.tir.schedule import Trace from ..builder import BuilderInput, BuilderResult, PyBuilder from ..mutator import PyMutator from ..runner import PyRunner, PyRunnerFuture, RunnerFuture, RunnerInput, RunnerResult from ..tune_context import TuneContext # pylint: disable=unused-import from ..utils import derived_object @derived_object class DummyRunnerFuture(PyRunnerFuture): def done(self) -> bool: return True def result(self) -> RunnerResult: run_secs = [random.uniform(5, 30) for _ in range(random.randint(1, 10))] return RunnerResult(run_secs, None) @derived_object class DummyBuilder(PyBuilder): def build(self, build_inputs: List[BuilderInput]) -> List[BuilderResult]: return [BuilderResult("test_path", None) for _ in build_inputs] @derived_object class DummyRunner(PyRunner): def run(self, runner_inputs: List[RunnerInput]) -> List[RunnerFuture]: return [DummyRunnerFuture() for _ in runner_inputs] # type: ignore @derived_object class DummyMutator(PyMutator): """Dummy Mutator for testing""" def _initialize_with_tune_context(self, context: "TuneContext") -> None: pass def apply(self, trace: Trace, _) -> Optional[Trace]: return Trace(trace.insts, {}) def clone(self): return DummyMutator()
2,182
33.109375
86
py
tvm
tvm-main/python/tvm/meta_schedule/testing/validate_database.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """JSON Database validation script""" import argparse import logging import warnings import itertools from statistics import mean from distutils.util import strtobool from typing import Callable, Tuple, Union, List, Any import numpy as np # type: ignore import tvm from tvm import meta_schedule as ms from tvm._ffi import get_global_func, register_func from tvm.ir import IRModule from tvm.support import describe from tvm.target import Target from tvm.tir import Schedule from tvm.tir.schedule import Trace from tvm.meta_schedule.utils import remove_build_dir from tvm.meta_schedule.testing.tune_utils import generate_input_data from tvm.tir.tensor_intrin import * # type: ignore # pylint: disable=wildcard-import,unused-wildcard-import DELIMITOR = "\n" + "-" * 30 + "\n" def _parse_args(): args = argparse.ArgumentParser() args.add_argument( "--work-dir", type=str, required=True, help="The path to the work directory containing database files.", ) args.add_argument( "--target", type=Target, required=True, ) args.add_argument( "--baseline-target", type=Target, default="llvm -num-cores=1", required=False, help="The baseline target to compile the original module.", ) args.add_argument( "--top-k", type=int, default=10**9, required=False, help="The number of top-k tuning records to validate for each unique original workload.", ) args.add_argument( "--rpc-host", type=str, ) args.add_argument( "--rpc-port", type=int, ) args.add_argument( "--rpc-key", type=str, ) args.add_argument( "--number", type=int, default=3, ) args.add_argument( "--repeat", type=int, default=1, ) args.add_argument( "--min-repeat-ms", type=int, default=100, ) args.add_argument( "--cpu-flush", type=lambda x: bool(strtobool(x)), help="example: True / False", required=True, ) args.add_argument( "--input-generator-func", type=str, default="tvm.meta_schedule.testing.default_input_generator", ) args.add_argument( "--check-metric-func", type=str, default="tvm.meta_schedule.testing.default_check_metric", ) parsed = args.parse_args() parsed.target = tvm.target.Target(parsed.target) if parsed.rpc_host is not None and parsed.rpc_port is not None and parsed.rpc_key is not None: parsed.rpc_config = ms.runner.RPCConfig( tracker_host=parsed.rpc_host, tracker_port=parsed.rpc_port, tracker_key=parsed.rpc_key, session_timeout_sec=600, ) else: parsed.rpc_config = None warnings.warn("RPC config is not provided, will use local runner.") if parsed.cpu_flush and parsed.target.kind.name != "llvm": warnings.warn("cpu_flush is only supported on llvm target") return parsed # arg parser ARGS = _parse_args() # logging logging.basicConfig( format="%(asctime)s.%(msecs)03d %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S" ) logging.getLogger("tvm.meta_schedule").setLevel(logging.DEBUG) logging.getLogger("tvm.meta_schedule.runner").setLevel(logging.WARN) def get_device_type(target: Target) -> str: """Get the device type string from a target. Parameters ---------- target : Target The target to get the device type from. Returns ------- device_type : str The device type string. """ if target.kind.name == "llvm": return "cpu" elif target.kind.name == "cuda": return "cuda" else: raise RuntimeError(f"Unsupported target kind for device type: {target.kind.name}") def get_runtime_device(target: Target) -> tvm.runtime.Device: """Get the runtime device from a target. Parameters ---------- target : Target The target to get the runtime device from. Returns ------- device : tvm.runtime.Device The runtime device. """ if target.kind.name == "llvm": return tvm.cpu() elif target.kind.name == "cuda": return tvm.cuda() else: raise RuntimeError(f"Unsupported target kind for runtime device: {target.kind.name}") def check_and_run(func: Union[str, Callable], *args, **kwargs) -> Any: """Check if the function is a string or a callable, and run it.""" if isinstance(func, str): func = get_global_func(func) return func(*args, **kwargs) # type: ignore class OriginalModule: """Original module class for deduplication.""" def __init__(self, mod: IRModule): self.mod = mod def __eq__(self, __o: "OriginalModule") -> bool: # type: ignore return tvm.ir.structural_equal(self.mod, __o.mod) def __hash__(self) -> int: return tvm.ir.structural_hash(self.mod) def initializer() -> None: """Initializer function to register the functions on PopenWorker.""" @register_func("tvm.meta_schedule.testing.default_check_metric") def default_check_metric( # pylint: disable=unused-variable,unreachable-code lhs: List[tvm.nd.NDArray], rhs: List[tvm.nd.NDArray] ) -> bool: """Check if the outputs are equal Parameters ---------- lhs : List[tvm.nd.NDArray] The first list of NDArrays to compare. rhs : List[tvm.nd.NDArray] The second list of NDArrays to compare. Returns ------- is_equal : bool Whether the two lists of NDArrays are equal. """ assert len(lhs) == len(rhs), "Different number of outputs from two modules" for i in range(len(lhs)): # pylint: disable=consider-using-enumerate if not np.allclose(lhs[i].numpy(), rhs[i].numpy(), rtol=1e-3, atol=2e-3): return False return True @register_func("tvm.meta_schedule.testing.default_input_generator") def default_input_generator( # pylint: disable=unused-variable mod: IRModule, ) -> List[tvm.nd.NDArray]: """Default input generator function Parameters ---------- mod : IRModule The IRModule to generate the input data for. Returns ------- inputs : List[tvm.nd.NDArray] The generated input data. """ args_info = ms.arg_info.TensorInfo.from_prim_func(mod["main"]) inputs = [ tvm.nd.array(generate_input_data(input_shape=arg_info.shape, input_dtype=arg_info.dtype)) for arg_info in args_info ] return inputs def to_numpy(a: List[tvm.nd.NDArray]) -> List[np.ndarray]: """Convert a list of TVM NDArray to a list of numpy array Parameters ---------- a : List[tvm.nd.NDArray] The list of TVM NDArray to be converted Returns ------- b : List[np.ndarray] The list of numpy array """ assert a is not None, "Empty result cannot be converted to numpy" return [x.numpy() for x in a] def to_tvm_ndarray(a: List[np.ndarray]) -> List[tvm.nd.NDArray]: """Convert a list of numpy array to a list of TVM NDArray Parameters ---------- a : List[np.ndarray] The list of numpy array to be converted. Returns ------- b : List[tvm.nd.NDArray] The list of TVM NDArray. """ assert a is not None, "Empty result cannot be converted to TVM NDArray" return [tvm.nd.array(x) for x in a] def is_failed_record(record: ms.database.TuningRecord) -> bool: """Check if a tuning record is failed. Parameters ---------- record : TuningRecord The tuning record to check. Returns ------- is_failed : bool """ return len(record.run_secs) == 1 and record.run_secs[0] == 1e9 def print_with_counter_func(counter: int, total: int) -> Callable: """Print with counter Parameters ---------- counter : int The counter to print with. total : int The total number of items to print with. Returns ------- print_result : Callable The print result function. """ def print_result( result: str, *, original_mod: IRModule = None, scheduled_mod: IRModule = None, inputs: List[np.ndarray] = None, original_res: List[np.ndarray] = None, scheduled_res: List[np.ndarray] = None, original_run_secs: List[float] = None, scheduled_run_secs: List[float] = None, exception: Exception = None, trace: str = None, ) -> None: """Print the validation result.""" status = f"Progress {counter: 6d} / {total: 6d} (estimated) checked, result: {result:>10}, " if result in ["pass", "wrong answer"]: status += ( f"original: {mean(original_run_secs) * 1e3: 10.3f} ms, " f"scheduled: {mean(scheduled_run_secs) * 1e3: 10.3f} ms" ) output = [status] if result not in ["pass", "skip"]: output.extend( [ "Original IRModule:" + DELIMITOR + original_mod.script(), "Scheduled IRModule:" + DELIMITOR + scheduled_mod.script(), "Trace" + DELIMITOR + str(trace), ] ) if result == "wrong answer": output.extend( [ "Input:" + DELIMITOR + str(inputs), "Original Result:" + DELIMITOR + str(original_res), "Scheduled Result:" + DELIMITOR + str(scheduled_res), "Max Diff:" + DELIMITOR + str( [ np.max(np.abs(original_res[i] - scheduled_res[i])) for i in range(len(original_res)) ] ) + "\n", ] ) elif result == "exception": output.extend(["Exception:" + DELIMITOR + str(exception) + "\n"]) else: raise ValueError(f"Unknown result: {result}") print("\n\n".join(output)) return print_result def make_alloc_arg_and_check( inputs: List[np.ndarray], original_mod: IRModule, scheduled_mod: IRModule, trace: str, original_res: List[np.ndarray], original_run_secs: List[float], print_result: Callable, ) -> Tuple[Callable, Callable]: """Make alloc_arg and check functions for the given inputs and collect results. Parameters ---------- inputs : List[np.ndarray] The inputs to the two modules. original_mod : IRModule The original IRModule. scheduled_mod : IRModule The scheduled IRModule. trace : str The trace of the scheduled IRModule. original_res : List[np.ndarray] The original results. original_run_secs : List[float] The original run times. print_result : Callable The print result function. Returns ------- f_with_args_alloc_argument : Callable The function to allocate arguments. f_with_args_run_evaluator : Callable The function to run evaluator. """ def f_with_args_alloc_argument_common( device: tvm.runtime.Device, args_info: ms.runner.rpc_runner.T_ARG_INFO_JSON_OBJ_LIST, # pylint: disable=unused-argument alloc_repeat: int, ) -> List[ms.runner.rpc_runner.T_ARGUMENT_LIST]: """Allocate arguments using the given inputs. Parameters ---------- session : RPCSession The RPC session. device : Device The device. args_info : T_ARG_INFO_JSON_OBJ_LIST argument information. alloc_repeat : int The number of times to repeat the allocation. Returns ------- args_list : List[T_ARGUMENT_LIST] The list of argument lists. """ return [[tvm.nd.array(arg, device=device) for arg in inputs] for _ in range(alloc_repeat)] def f_with_args_run_evaluator_common( rt_mod: tvm.runtime.Module, device: tvm.runtime.Device, evaluator_config: ms.runner.EvaluatorConfig, repeated_args: List[ms.runner.rpc_runner.T_ARGUMENT_LIST], ) -> List[float]: """With args function to run the evaluator Parameters ---------- session : tvm.rpc.RPCSession The RPC session rt_mod: Module The runtime module device: Device The device to run the evaluator evaluator_config: EvaluatorConfig The evaluator config repeated_args: List[T_ARGUMENT_LIST] The repeated arguments Returns ------- costs: List[float] The evaluator results """ evaluator = rt_mod.time_evaluator( func_name=rt_mod.entry_name, dev=device, number=evaluator_config.number, repeat=evaluator_config.repeat, min_repeat_ms=evaluator_config.min_repeat_ms, f_preproc="cache_flush_cpu_non_first_arg" if evaluator_config.enable_cpu_cache_flush else "", ) repeated_costs: List[List[float]] = [] for args in repeated_args: device.sync() profile_result = evaluator(*args) repeated_costs.append(profile_result.results) costs = [float(cost) for cost in itertools.chain.from_iterable(repeated_costs)] assert len(repeated_args) == 1, "Only support one set of arguments" scheduled_res = [arg.numpy() for arg in repeated_args[0]] # type: ignore # fetch comparison function passed = check_and_run( ARGS.check_metric_func, to_tvm_ndarray(original_res), to_tvm_ndarray(scheduled_res), ) print_result( result="pass" if passed else "wrong answer", original_mod=original_mod, scheduled_mod=scheduled_mod, trace=trace, inputs=inputs, original_res=original_res, scheduled_res=scheduled_res, original_run_secs=original_run_secs, scheduled_run_secs=costs, ) return costs def f_with_args_alloc_argument_rpc( rpc_session: ms.runner.rpc_runner.RPCSession, # pylint: disable=unused-argument device: tvm.runtime.Device, args_info: ms.runner.rpc_runner.T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[ms.runner.rpc_runner.T_ARGUMENT_LIST]: return f_with_args_alloc_argument_common(device, args_info, alloc_repeat) def f_with_args_run_evaluator_rpc( rpc_session: ms.runner.rpc_runner.RPCSession, # pylint: disable=unused-argument rt_mod: tvm.runtime.Module, device: tvm.runtime.Device, evaluator_config: ms.runner.EvaluatorConfig, repeated_args: List[ms.runner.rpc_runner.T_ARGUMENT_LIST], ) -> List[float]: return f_with_args_run_evaluator_common(rt_mod, device, evaluator_config, repeated_args) if ARGS.rpc_config is None: return f_with_args_alloc_argument_common, f_with_args_run_evaluator_common else: return f_with_args_alloc_argument_rpc, f_with_args_run_evaluator_rpc def local_build_and_run( mod: IRModule, target: Target, device: tvm.runtime.Device, inputs: List[np.ndarray], ) -> Tuple[List[np.ndarray], List[float]]: """Build and run the module locally. Parameters ---------- mod: IRModule The module to build and run target: Target The target to build the module device: Device The device to run the module inputs: List[np.ndarray] The inputs to run the module Returns ------- res: List[np.ndarray] The results of running the module run_secs: List[float] The running time of running the module """ # potential memory leak https://github.com/apache/tvm/issues/11096 lib = tvm.build(mod, target=target) tvm_inputs = [tvm.nd.array(inp, device=device) for inp in inputs] device.sync() func = lib.time_evaluator(lib.entry_name, dev=device, number=ARGS.number, repeat=ARGS.repeat) benchmark_res = func(*tvm_inputs) device.sync() return [arg.numpy() for arg in tvm_inputs], list(benchmark_res.results) def _check_builder_result(builder_result: ms.builder.BuilderResult) -> None: """Check if the builder result is defined. Parameters ---------- builder_result: BuilderResult The builder result """ assert builder_result.error_msg is None, "Builder failed: " + str( builder_result.error_msg if builder_result.error_msg else "Empty error message" ) def _apply_trace(mod: IRModule, trace: Trace) -> IRModule: """Apply the trace to the module. Parameters ---------- mod: IRModule The module to apply the trace to trace: Trace The trace to apply Returns ------- mod: IRModule The module with the trace applied """ sch = Schedule(mod) trace.apply_to_schedule(sch, remove_postproc=False) return sch.mod def _build_all_mods( mods: List[IRModule], builder: ms.builder.Builder, target: Target ) -> List[ms.builder.BuilderResult]: """Build all the modules. Parameters ---------- mods: List[IRModule] The modules to build builder: Builder The builder to build the modules target: Target The target to build the modules Returns ------- builder_results: List[BuilderResult] The builder results """ builder_results = builder.build([ms.builder.BuilderInput(mod, target) for mod in mods]) assert len(builder_results) == len( mods ), f"Unexpected number of build results, expected {len(mods)} got {len(builder_results)}" return builder_results def _run_single_mod( builder_result: ms.builder.BuilderResult, runner: ms.runner.Runner, dev_type: str, ) -> None: """Run a single module. Parameters ---------- builder_result: BuilderResult The builder result runner: Runner The runner to run the module dev_type: str The device type """ runner_futures = runner.run( # arginfo is not used in this case so we can pass an empty list [ms.runner.RunnerInput(builder_result.artifact_path, device_type=dev_type, args_info=[])] ) assert ( len(runner_futures) == 1 ), f"Unexpected number of runner futures, expected 1 got {len(runner_futures)}" (runner_future,) = runner_futures # pylint: disable=unbalanced-tuple-unpacking runner_res = runner_future.result() assert runner_res.error_msg is None, "Runner failed: " + ( runner_res.error_msg if runner_res.error_msg else "Empty error message" ) def main(): """Main function""" describe() with ms.Profiler() as profiler: # initialize target = ARGS.target dev_type = get_device_type(target) builder = ms.builder.LocalBuilder() database = ms.database.create(work_dir=ARGS.work_dir) # collect records with profiler.timeit("collect records"): records = database.get_all_tuning_records() total = len(records) print( f"Total {total} records to be validated. " f"Collected in {float(profiler.get()['collect records']): 3.3f} sec." ) # collect unique original TIR with profiler.timeit("deduplicate records"): workloads = set() for record in records: workloads.add(OriginalModule(record.workload.mod)) print( f"Total {len(workloads)} unique original TIR to validate. " f"Deduplicated in {float(profiler.get()['deduplicate records']): 3.3f} sec." ) if ARGS.top_k < 10**9: print(f"Top {ARGS.top_k} records for each original TIR will be validated.") total = len(workloads) * ARGS.top_k print() # validate correctness counter = 0 for item in workloads: original_mod = item.mod records = database.get_top_k( workload=database.commit_workload(original_mod), top_k=ARGS.top_k ) if len(records) < ARGS.top_k: total -= ARGS.top_k - len(records) inputs = to_numpy(check_and_run(ARGS.input_generator_func, original_mod)) original_res, original_run_secs = local_build_and_run( original_mod, target=ARGS.baseline_target, inputs=inputs, device=get_runtime_device(ARGS.baseline_target), ) scheduled_mods = [_apply_trace(original_mod, record.trace) for record in records] builder_results = _build_all_mods(scheduled_mods, builder, target) # type: ignore for i, record in enumerate(records): counter += 1 print_result = print_with_counter_func(counter=counter, total=total) if is_failed_record(record): # skip failed records where run_secs is 1e9 # these records are only negative samples for cost model print_result(result="skip") continue try: # prepare scheduled module scheduled_mod = scheduled_mods[i] # check build result builder_result = builder_results[i] _check_builder_result(builder_result) # fetch functions ( f_with_args_alloc_argument, f_with_args_run_evaluator, ) = make_alloc_arg_and_check( inputs, original_mod, scheduled_mod, str(record.trace), original_res=original_res, original_run_secs=original_run_secs, print_result=print_result, ) # create runner evaluator_config = ms.runner.EvaluatorConfig( number=ARGS.number, repeat=ARGS.repeat, min_repeat_ms=ARGS.min_repeat_ms, enable_cpu_cache_flush=ARGS.cpu_flush, ) if ARGS.rpc_config is not None: runner: ms.Runner = ms.runner.RPCRunner( # type: ignore ARGS.rpc_config, evaluator_config=evaluator_config, alloc_repeat=1, f_alloc_argument=f_with_args_alloc_argument, f_run_evaluator=f_with_args_run_evaluator, initializer=initializer, ) else: runner: ms.Runner = ms.runner.LocalRunner( # type: ignore evaluator_config=evaluator_config, alloc_repeat=1, f_alloc_argument=f_with_args_alloc_argument, f_run_evaluator=f_with_args_run_evaluator, initializer=initializer, ) # run and validate _run_single_mod(builder_result, runner, dev_type) # type: ignore except Exception as e: # pylint: disable=broad-except, invalid-name # validation failed with exception print_result( result="exception", original_mod=original_mod, scheduled_mod=scheduled_mod, trace=str(record.trace), exception=e, ) # clean up remove_build_dir(builder_result.artifact_path) print(f"Validation finished! Total time spent: {float(profiler.get()['Total']): 3.3f} sec.") if __name__ == "__main__": main()
25,351
31.670103
108
py
tvm
tvm-main/python/tvm/meta_schedule/testing/dataset_sample_candidates.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import glob import json import os from typing import List from tqdm import tqdm # type: ignore import tvm from tvm import meta_schedule as ms from tvm.ir import load_json from tvm.target import Target def _parse_args(): parser = argparse.ArgumentParser() parser.add_argument( "--task_cache_dir", type=str, help="Please provide the full path to the extracted tasks." ) parser.add_argument( "--candidate_cache_dir", type=str, help="Please provide the full path to save the sampled candidates.", ) parser.add_argument( "--target", type=str, default="nvidia/geforce-rtx-3070", help="Please specify the target hardware for tuning.\ Note: for generating dataset, the hardware does not need to be present.", ) parser.add_argument( "--init_population_size", type=int, default=256, help="The initial population size used in evolutionary search.", ) parser.add_argument( "--num_samples_per_task", type=int, default=400, help="The number of samples to gather per tuning task.", ) parser.add_argument( "--num_trials_per_iter", type=int, default=64, help="The number of trials per iteration in evolutionary search.", ) parser.add_argument( "--max_trials_per_task", type=int, default=400, help="The maximum number of trials per task in evolutionary search.", ) parser.add_argument( "--max_retry_per_task", type=int, default=10, help="The maximum number of retry attempts allowed.", ) parser.add_argument( "--file_group", type=int, default=0, help="To enable running multiple scripts in parallel, files [idx * 10 : (idx + 1) * 10]\ in the sorted file list from the given directory will be run.", ) return parser.parse_args() # pylint: disable=too-many-locals def sample_candidates(task, task_name, model_name): """Randomly sample candidates for a task and save the candidates in the given directory. Parameters ---------- task : IRModule The initial ir module used for generating the search space. task_name : str The name of the task. model_name : str The name of the model. Returns ------- None """ candidate_path = os.path.join( args.candidate_cache_dir, model_name, task_name + "_candidates.json" ) workload_path = os.path.join(args.candidate_cache_dir, model_name, task_name + "_workload.json") database = ms.database.JSONDatabase( path_workload=workload_path, path_tuning_record=candidate_path, ) sample_init_population = tvm.get_global_func( "meta_schedule.SearchStrategyEvolutionarySearchSampleInitPopulation" ) evolve_with_cost_model = tvm.get_global_func( "meta_schedule.SearchStrategyEvolutionarySearchEvolveWithCostModel" ) strategy = ms.search_strategy.EvolutionarySearch(init_measured_ratio=0.0) target = Target(args.target) context = ms.TuneContext( mod=task, target=target, space_generator="post-order-apply", search_strategy=strategy, task_name=task_name, ) context.initialize() context.pre_tuning( max_trials=args.max_trials_per_task, num_trials_per_iter=args.num_trials_per_iter, design_spaces=context.generate_design_space(), database=database, cost_model=ms.cost_model.RandomModel(), # type: ignore ) all_states: List[tvm.tir.Schedule] = [] num_retry, itr = 0, 0 states = sample_init_population(strategy, args.init_population_size) while len(all_states) < args.num_samples_per_task and num_retry < args.max_retry_per_task: states = evolve_with_cost_model(strategy, states, len(states)) all_states += states if len(states) == 0: states = sample_init_population(strategy, args.init_population_size) num_retry += 1 else: num_retry = 0 print(f"iter: {itr}, number of states sampled: {len(all_states)}") itr += 1 all_states = all_states[: args.num_samples_per_task] workload = ms.database.Workload(context.mod) database.commit_workload(context.mod) for state in all_states: database.commit_tuning_record(ms.database.TuningRecord(state.trace, workload)) args = _parse_args() # pylint: disable=invalid-name def main(): if not os.path.isdir(args.task_cache_dir): raise Exception("Please provide a correct task cache dir.") try: os.makedirs(args.candidate_cache_dir, exist_ok=True) except OSError: print(f"Directory {args.candidate_cache_dir} cannot be created successfully.") task_paths = sorted(glob.glob(os.path.join(args.task_cache_dir, "*.json")))[ args.file_group * 10 : (args.file_group + 1) * 10 ] print(f"Selected models: {task_paths}") for num, task_path in enumerate(task_paths): print(f"Processing model {num} ...") with open(task_path, "rb") as file: tasks = file.readlines() model_name = task_path.split("/")[-1][len("relay-") :][: -len("_extracted_tasks.json")] os.makedirs(os.path.join(args.candidate_cache_dir, model_name), exist_ok=True) for task_str in tqdm(tasks): task_name, task_mod = json.loads(task_str) task_mod = load_json(json.dumps(task_mod)) sample_candidates(task_mod, task_name, model_name) if __name__ == "__main__": main()
6,498
33.569149
100
py
tvm
tvm-main/python/tvm/meta_schedule/testing/tlcbench.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name,import-outside-toplevel # type: ignore """Model loader for TLCBench.""" import logging import multiprocessing import os import tvm from tvm import relay from tvm.contrib.download import download_testdata log = logging.getLogger(__name__) def _convert(args): onnx_model, shape_dict, json_path, params_path = args mod, params = relay.frontend.from_onnx(onnx_model, shape_dict, freeze_params=True) seq = tvm.transform.Sequential( [relay.transform.InferType(), relay.transform.FakeQuantizationToInteger(use_qat=True)] ) mod = seq(mod) with open(json_path, "w") as fo: fo.write(tvm.ir.save_json(mod)) with open(params_path, "wb") as fo: fo.write(relay.save_param_dict(params)) def convert_to_qnn(onnx_path, json_path, params_path, input_info): """Run the ONNX frontend and the FQ2I pass. The output is serialized to disk.""" import onnx onnx_model = onnx.load(onnx_path) shape_dict = dict(input_info) log.info("Converting te ONNX model to Relay and running the FQ2I pass, it may take a while...") with multiprocessing.Pool(processes=1) as pool: pool.map(_convert, [(onnx_model, shape_dict, json_path, params_path)]) def deserialize_relay(json_path, params_path): with open(json_path, "r") as fi: mod = tvm.ir.load_json(fi.read()) with open(params_path, "rb") as fi: params = relay.load_param_dict(fi.read()) return mod, params def load_quantized_bert_base(batch_size=1, seq_len=384): """ Load the quantized bert-base model from TLCBench, possibly downloading it from github and caching the converted int8 QNN module to disk. In addition to returing the relay module and its parameters, it also returns input name and shape information, which can be used at the deployment time as follows: ``` mod, params, input_info = load_quantized_bert_base() ... runtime = tvm.contrib.graph_executor.GraphModule(lib["default"](dev)) for name, shape in input_info: arr = np.random.uniform(1, 10, size=shape).astype("int64") runtime.set_input(name, arr) runtime.run() ``` """ url = "https://github.com/tlc-pack/TLCBench/raw/main/models/bert-base-qat.onnx" log.info("Downloading quantized bert-base model.") onnx_path = download_testdata(url, "bert-base-qat.onnx", module="tlcbench") data_dir = os.path.dirname(onnx_path) json_path = os.path.join(data_dir, "bert_base_int8_b%d_s%d.json" % (batch_size, seq_len)) params_path = os.path.join(data_dir, "bert_base_int8_b%d_s%d.params" % (batch_size, seq_len)) # Input names and order encoded in the ONNX model input_info = [ ("input_ids", (batch_size, seq_len)), ("segment_ids", (batch_size, seq_len)), ("input_mask", (batch_size, seq_len)), ] if not os.path.exists(json_path) or not os.path.exists(params_path): convert_to_qnn(onnx_path, json_path, params_path, input_info) def deserialize(): try: return deserialize_relay(json_path, params_path) except ValueError: # A serialized Relay json file may become invalid after TVM bump # Update the serialized model and try loading again convert_to_qnn(onnx_path, json_path, params_path, input_info) return deserialize_relay(json_path, params_path) mod, params = deserialize() return mod, params, input_info
4,266
33.691057
99
py
tvm
tvm-main/python/tvm/meta_schedule/testing/dataset_extract_tasks.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import glob import json import os from tqdm import tqdm # type: ignore import tvm from tvm import meta_schedule as ms from tvm.ir import save_json from tvm.meta_schedule.testing.relay_workload import _load_cache from tvm.runtime import load_param_dict def _parse_args(): parser = argparse.ArgumentParser() parser.add_argument( "--model_cache_dir", type=str, help="Please provide the full path to the model cache dir." ) parser.add_argument( "--task_cache_dir", type=str, help="Please provide the full path to save extracted tasks." ) parser.add_argument( "--target", type=str, default="cuda", help="Please specify the target hardware for tuning." ) return parser.parse_args() # pylint: disable=too-many-locals def extract_and_save_tasks(cache_file): """Extract tuning tasks and cache the nonspatial ones in the given directory. Parameters ---------- cache_file : str The filename of the cached model. Returns ------- None """ mod, params_bytearray, _ = _load_cache(args.model_cache_dir, cache_file) params = load_param_dict(params_bytearray) try: extracted_tasks = ms.relay_integration.extract_tasks(mod, target=args.target, params=params) except tvm.error.TVMError as error: print(str(error)) return task_cache_path = os.path.join( args.task_cache_dir, cache_file.split(".")[0] + "_extracted_tasks.json" ) is_spatial = tvm.get_global_func("tir.schedule.IsSpatialPrimFunc") with open(task_cache_path, "w", encoding="utf8") as file: for i, task in enumerate(extracted_tasks): subgraph = task.dispatched[0] prim_func = subgraph[subgraph.get_global_vars()[0]] if not is_spatial(prim_func): subgraph_str = save_json(subgraph) json_obj = [task.task_name, json.loads(subgraph_str)] json_str = json.dumps(json_obj) assert "\n" not in json_str, "Failed to generate single line string." if i == len(extracted_tasks) - 1: file.write(json_str) else: file.write(json_str + "\n") args = _parse_args() # pylint: disable=invalid-name def main(): if not os.path.isdir(args.model_cache_dir): raise Exception("Please provide a correct model cache dir.") try: os.makedirs(args.task_cache_dir, exist_ok=True) except OSError: print(f"Directory {args.task_cache_dir} cannot be created successfully.") paths = glob.glob(os.path.join(args.model_cache_dir, "*.json")) # pylint: disable=invalid-name for path in tqdm(paths): filename = path.split("/")[-1] extract_and_save_tasks(filename) if __name__ == "__main__": main()
3,657
33.838095
100
py
tvm
tvm-main/python/tvm/meta_schedule/testing/tune_te.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import logging from distutils.util import strtobool from typing import Optional import tvm from tvm import meta_schedule as ms from tvm import tir from tvm.meta_schedule.testing.te_workload import create_te_workload from tvm.support import describe def _parse_args(): args = argparse.ArgumentParser() args.add_argument( "--workload", type=str, required=True, ) args.add_argument( "--target", type=str, required=True, ) args.add_argument( "--num-trials", type=int, required=True, ) args.add_argument( "--rpc-host", type=str, required=True, ) args.add_argument( "--rpc-port", type=int, required=True, ) args.add_argument( "--rpc-key", type=str, required=True, ) args.add_argument( "--work-dir", type=str, required=True, ) args.add_argument( "--number", type=int, default=3, ) args.add_argument( "--repeat", type=int, default=1, ) args.add_argument( "--min-repeat-ms", type=int, default=100, ) args.add_argument( "--adaptive-training", type=lambda x: bool(strtobool(x)), required=False, help="example: True / False", default=True, ) args.add_argument( "--cpu-flush", type=lambda x: bool(strtobool(x)), help="example: True / False", required=True, ) parsed = args.parse_args() parsed.target = tvm.target.Target(parsed.target) parsed.rpc_config = ms.runner.RPCConfig( tracker_host=parsed.rpc_host, tracker_port=parsed.rpc_port, tracker_key=parsed.rpc_key, session_timeout_sec=60, ) return parsed logging.basicConfig( format="%(asctime)s.%(msecs)03d %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S" ) logging.getLogger("tvm.meta_schedule").setLevel(logging.DEBUG) ARGS = _parse_args() def main(): describe() print(f"Workload: {ARGS.workload}") with ms.Profiler() as profiler: sch: Optional[tir.Schedule] = ms.tir_integration.tune_tir( mod=create_te_workload(ARGS.workload, 0), target=ARGS.target, work_dir=ARGS.work_dir, max_trials_global=ARGS.num_trials, num_trials_per_iter=64, runner=ms.runner.RPCRunner( # type: ignore rpc_config=ARGS.rpc_config, evaluator_config=ms.runner.EvaluatorConfig( number=ARGS.number, repeat=ARGS.repeat, min_repeat_ms=ARGS.min_repeat_ms, enable_cpu_cache_flush=ARGS.cpu_flush, ), alloc_repeat=1, ), cost_model=ms.cost_model.XGBModel( # type: ignore extractor=ms.feature_extractor.PerStoreFeature(), adaptive_training=ARGS.adaptive_training, ), strategy=ms.search_strategy.EvolutionarySearch(), ) print("Tuning Time:") print(profiler.table()) if sch is None: print("No valid schedule found!") else: print(sch.mod.script()) print(sch.trace) if __name__ == "__main__": main()
4,191
26.578947
91
py
tvm
tvm-main/python/tvm/meta_schedule/testing/te_workload.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Workloads in TE""" # pylint: disable=missing-docstring from typing import Tuple from tvm import te, tir, topi from tvm.target import Target def batch_matmul_nkkm( # pylint: disable=invalid-name,missing-docstring B: int, N: int, M: int, K: int, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: x = te.placeholder((B, N, K), name="X", dtype=in_dtype) y = te.placeholder((B, K, M), name="Y", dtype=in_dtype) k = te.reduce_axis((0, K), name="k") z = te.compute( # pylint: disable=invalid-name (B, N, M), lambda b, i, j: te.sum( x[b][i][k].astype(out_dtype) * y[b][k][j].astype(out_dtype), axis=[k], ), name="Z", ) return (x, y, z) def conv1d_nlc( # pylint: disable=invalid-name,missing-docstring N: int, L: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, dilation: int = 1, groups: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder((N, L, CI), name="inputs", dtype=in_dtype) weight = te.placeholder((kernel_size, CI // groups, CO), name="weight", dtype=in_dtype) batch_size, in_len, _ = inputs.shape k_len, channel_per_group, out_channel = weight.shape out_channel_per_group = out_channel // groups out_len = (in_len + 2 * padding - dilation * (k_len - 1) - 1) // stride + 1 rc = te.reduce_axis((0, channel_per_group), name="rc") rl = te.reduce_axis((0, k_len), name="rl") padded = topi.nn.pad(inputs, [0, padding, 0]) output = te.compute( (batch_size, out_len, out_channel), lambda n, l, co: te.sum( ( padded[ n, l * stride + rl * dilation, co // out_channel_per_group * channel_per_group + rc, ].astype(out_dtype) * weight[rl, rc, co].astype(out_dtype) ), axis=[rl, rc], ), name="conv1d_nlc", ) return (inputs, weight, output) def conv2d_nhwc( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, dilation: int = 1, groups: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder((N, H, W, CI), name="inputs", dtype=in_dtype) weight = te.placeholder( (kernel_size, kernel_size, CI // groups, CO), name="weight", dtype=in_dtype ) batch_size, in_h, in_w, _ = inputs.shape k_h, k_w, channel_per_group, out_channel = weight.shape out_channel_per_group = out_channel // groups out_h = (in_h + 2 * padding - dilation * (k_h - 1) - 1) // stride + 1 out_w = (in_w + 2 * padding - dilation * (k_w - 1) - 1) // stride + 1 rh = te.reduce_axis((0, k_h), name="rh") rw = te.reduce_axis((0, k_w), name="rw") rc = te.reduce_axis((0, channel_per_group), name="rc") padded = topi.nn.pad(inputs, [0, padding, padding, 0]) output = te.compute( (batch_size, out_h, out_w, out_channel), lambda n, h, w, co: te.sum( ( padded[ n, h * stride + rh * dilation, w * stride + rw * dilation, co // out_channel_per_group * channel_per_group + rc, ].astype(out_dtype) * weight[rh, rw, rc, co].astype(out_dtype) ), axis=[rh, rw, rc], ), name="conv2d_nhwc", ) return (inputs, weight, output) def conv3d_ndhwc( # pylint: disable=invalid-name,missing-docstring N: int, D: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, dilation: int = 1, groups: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder((N, D, H, W, CI), name="inputs", dtype=in_dtype) weight = te.placeholder( (kernel_size, kernel_size, kernel_size, CI // groups, CO), name="weight", dtype=in_dtype ) batch_size, in_d, in_h, in_w, _ = inputs.shape k_d, k_h, k_w, channel_per_group, out_channel = weight.shape out_channel_per_group = out_channel // groups out_d = (in_d + 2 * padding - dilation * (k_d - 1) - 1) // stride + 1 out_h = (in_h + 2 * padding - dilation * (k_h - 1) - 1) // stride + 1 out_w = (in_w + 2 * padding - dilation * (k_w - 1) - 1) // stride + 1 rd = te.reduce_axis((0, k_d), name="rd") rh = te.reduce_axis((0, k_h), name="rh") rw = te.reduce_axis((0, k_w), name="rw") rc = te.reduce_axis((0, channel_per_group), name="rc") padded = topi.nn.pad(inputs, [0, padding, padding, padding, 0]) output = te.compute( (batch_size, out_d, out_h, out_w, out_channel), lambda n, d, h, w, co: te.sum( ( padded[ n, d * stride + rd * dilation, h * stride + rh * dilation, w * stride + rw * dilation, co // out_channel_per_group * channel_per_group + rc, ].astype(out_dtype) * weight[rd, rh, rw, rc, co].astype(out_dtype) ), axis=[rd, rh, rw, rc], ), name="conv3d_ndhwc", ) return (inputs, weight, output) def depthwise_conv2d_nhwc( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, C: int, kernel_size: int, stride: int = 1, padding: int = 0, dilation: int = 1, factor: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder((N, H, W, C), dtype=in_dtype) weight = te.placeholder((factor, kernel_size, kernel_size, C), dtype=in_dtype) batch_size, in_h, in_w, in_channel = inputs.shape factor, k_h, k_w, in_channel = weight.shape out_channel = in_channel * factor assert int(factor) == 1, "Not optimized for factor != 1" out_h = (in_h + 2 * padding - dilation * (k_h - 1) - 1) // stride + 1 out_w = (in_w + 2 * padding - dilation * (k_w - 1) - 1) // stride + 1 rh = te.reduce_axis((0, k_h), name="rh") rw = te.reduce_axis((0, k_w), name="rw") padded = topi.nn.pad(inputs, [0, padding, padding, 0]) output = te.compute( (batch_size, out_h, out_w, out_channel), lambda n, h, w, c: te.sum( ( padded[ n, h * stride + rh * dilation, w * stride + rw * dilation, c // factor, ].astype(out_dtype) * weight[c % factor, rh, rw, c // factor].astype(out_dtype) ), axis=[rh, rw], ), name="depth_conv2d_nhwc", ) return (inputs, weight, output) def conv2d_transpose_nhwc( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder((N, H, W, CI), name="inputs", dtype=in_dtype) weight = te.placeholder((kernel_size, kernel_size, CI, CO), name="weight", dtype=in_dtype) batch, in_h, in_w, in_c = inputs.shape filter_h, filter_w, in_c, out_c = weight.shape stride_h, stride_w = (stride, stride) # compute padding fpad_top, fpad_left, fpad_bottom, fpad_right = topi.nn.get_pad_tuple( padding, (filter_h, filter_w) ) bpad_top = filter_h - 1 - fpad_top bpad_bottom = filter_h - 1 - fpad_bottom bpad_left = filter_w - 1 - fpad_left bpad_right = filter_w - 1 - fpad_right # padding stage padded = topi.nn.pad( inputs, [ 0, (bpad_top + stride_h - 1) // stride_h, (bpad_left + stride_w - 1) // stride_w, 0, ], [ 0, (bpad_bottom + stride_h - 1) // stride_h, (bpad_right + stride_w - 1) // stride_w, 0, ], ) # remove extra padding introduced by dilatation idx_div = te.indexdiv idx_mod = te.indexmod border_h = idx_mod(stride_h - idx_mod(bpad_top, stride_h), stride_h) border_w = idx_mod(stride_w - idx_mod(bpad_left, stride_w), stride_w) # dilation stage strides = [1, stride_h, stride_w, 1] n = len(padded.shape) # We should embed this dilation directly into te.compute rather than creating a new te.compute. # Only in this way can we use unroll to eliminate the multiplication of zeros. def _dilate(*indices): not_zero = [] index_tuple = [] for i in range(n): if not strides[i] == 1: index_tuple.append(idx_div(indices[i], strides[i])) not_zero.append(idx_mod(indices[i], strides[i]).equal(0)) else: index_tuple.append(indices[i]) if not_zero: not_zero = te.all(*not_zero) return te.if_then_else(not_zero, padded(*index_tuple), tir.const(0.0, padded.dtype)) return padded(*index_tuple) # convolution stage out_h = (in_h - 1) * stride_h - fpad_top - fpad_bottom + filter_h out_w = (in_w - 1) * stride_w - fpad_left - fpad_right + filter_w rc = te.reduce_axis((0, in_c), name="rc") rh = te.reduce_axis((0, filter_h), name="rh") rw = te.reduce_axis((0, filter_w), name="rw") output = te.compute( (batch, out_h, out_w, out_c), lambda n, h, w, co: te.sum( _dilate(n, h + rh + border_h, w + rw + border_w, rc).astype(out_dtype) * weight[filter_h - 1 - rh, filter_w - 1 - rw, rc, co].astype(out_dtype), axis=[rh, rw, rc], ), name="conv2d_transpose_nhwc", ) return (inputs, weight, output) def conv2d_capsule_nhwijc( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, capsule_size: int = 4, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: inputs = te.placeholder( (N, H, W, capsule_size, capsule_size, CI), name="inputs", dtype=in_dtype ) weight = te.placeholder( (kernel_size, kernel_size, capsule_size, capsule_size, CI, CO), name="weight", dtype=in_dtype, ) batch_size, in_h, in_w, _, _, in_channel = inputs.shape k_h, k_w, _, _, _, out_channel = weight.shape out_h = (in_h + 2 * padding - kernel_size) // stride + 1 out_w = (in_w + 2 * padding - kernel_size) // stride + 1 rh = te.reduce_axis((0, k_h), name="rh") rw = te.reduce_axis((0, k_w), name="rw") cap_k = te.reduce_axis((0, capsule_size), name="cap_k") rc = te.reduce_axis((0, in_channel), name="rc") padded = topi.nn.pad(inputs, [0, padding, padding, 0, 0, 0]) output = te.compute( (batch_size, out_h, out_w, capsule_size, capsule_size, out_channel), lambda n, h, w, cap_i, cap_j, co: te.sum( ( padded[n, h * stride + rh, w * stride + rw, cap_i, cap_k, rc].astype(out_dtype) * weight[rh, rw, cap_k, cap_j, rc, co].astype(out_dtype) ), axis=[rh, rw, cap_k, rc], ), name="conv2d_capsule_nhwijc", ) return (inputs, weight, output) def norm_bmn( # pylint: disable=invalid-name,missing-docstring B: int, M: int, N: int, ) -> Tuple[te.Tensor, te.Tensor]: a = te.placeholder((B, M, N), name="A") i = te.reduce_axis((0, M), name="i") j = te.reduce_axis((0, N), name="j") c = te.compute( (B,), lambda b: te.sum(a[b][i][j] * a[b][i][j], axis=[i, j]), name="C", ) d = te.compute((B,), lambda b: te.sqrt(c[b]), name="D") return (a, d) def conv2d_nhwc_without_layout_rewrite( # pylint: disable=invalid-name Input: te.Tensor, Filter: te.Tensor, stride: int, padding: int, dilation: int, out_dtype="float32", ): """A copy of `topi.nn.conv2d_nhwc` but without the 'layout_free` attribute. We use this in single op and subgraph evaluation because we don't want to introduce graph level optimization. """ assert isinstance(stride, int) or len(stride) == 2 assert isinstance(dilation, int) or len(dilation) == 2 if isinstance(stride, int): stride_h = stride_w = stride else: stride_h, stride_w = stride if isinstance(dilation, int): dilation_h = dilation_w = dilation else: dilation_h, dilation_w = dilation batch, in_height, in_width, in_channel = Input.shape # type: ignore kernel_h, kernel_w, _channel, num_filter = Filter.shape # type: ignore # compute the output shape dilated_kernel_h = (kernel_h - 1) * dilation_h + 1 dilated_kernel_w = (kernel_w - 1) * dilation_w + 1 pad_top, pad_left, pad_down, pad_right = topi.nn.get_pad_tuple( padding, (dilated_kernel_h, dilated_kernel_w) ) out_channel = num_filter out_height = topi.utils.simplify( (in_height - dilated_kernel_h + pad_top + pad_down) // stride_h + 1 ) out_width = topi.utils.simplify( (in_width - dilated_kernel_w + pad_left + pad_right) // stride_w + 1 ) pad_before = [0, pad_top, pad_left, 0] pad_after = [0, pad_down, pad_right, 0] PaddedInput = topi.nn.pad(Input, pad_before, pad_after, name="PaddedInput") rc = te.reduce_axis((0, in_channel), name="rc") ry = te.reduce_axis((0, kernel_h), name="ry") rx = te.reduce_axis((0, kernel_w), name="rx") Output = te.compute( (batch, out_height, out_width, out_channel), lambda nn, yy, xx, ff: te.sum( PaddedInput[ nn, yy * stride_h + ry * dilation_h, xx * stride_w + rx * dilation_w, rc ].astype(out_dtype) * Filter[ry, rx, rc, ff].astype(out_dtype), # type: ignore axis=[ry, rx, rc], ), name="Conv2dOutput", tag="conv2d_nhwc", ) return Output def conv2d_nhwc_bn_relu( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, strides: int, padding: int, dilation: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor, te.Tensor, te.Tensor, te.Tensor]: data = te.placeholder((N, H, W, CI), name="data", dtype=in_dtype) kernel = te.placeholder((kernel_size, kernel_size, CI, CO), name="kernel", dtype=in_dtype) bias = te.placeholder((CO,), name="bias") bn_scale = te.placeholder((CO,), name="bn_scale") bn_offset = te.placeholder((CO,), name="bn_offset") OH = (H + 2 * padding - (kernel_size - 1) * dilation - 1) // strides + 1 OW = (W + 2 * padding - (kernel_size - 1) * dilation - 1) // strides + 1 conv = conv2d_nhwc_without_layout_rewrite(data, kernel, strides, padding, dilation, out_dtype) conv = te.compute( (N, OH, OW, CO), lambda i, j, k, l: conv[i, j, k, l] + bias[l], name="bias_add" ) conv = te.compute( (N, OH, OW, CO), lambda i, j, k, l: conv[i, j, k, l] * bn_scale[l], name="bn_mul" ) conv = te.compute( (N, OH, OW, CO), lambda i, j, k, l: conv[i, j, k, l] + bn_offset[l], name="bn_add" ) out = topi.nn.relu(conv) return (data, kernel, bias, bn_offset, bn_scale, out) def transpose_batch_matmul( # pylint: disable=invalid-name,missing-docstring batch: int, seq_len: int, n_head: int, n_dim: int, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: query = te.placeholder((batch, seq_len, n_head, n_dim), name="query", dtype=in_dtype) value = te.placeholder((batch, seq_len, n_head, n_dim), name="value", dtype=in_dtype) query_T = te.compute( (batch, n_head, seq_len, n_dim), lambda b, h, l, d: query[b, l, h, d], name="query_T", ) value_T = te.compute( (batch, n_head, n_dim, seq_len), lambda b, h, d, l: value[b, l, h, d], name="value_T", ) k = te.reduce_axis((0, n_dim), name="k") out = te.compute( (batch, n_head, seq_len, seq_len), lambda b, h, i, j: te.sum( query_T[b, h, i, k].astype(out_dtype) * value_T[b, h, k, j].astype(out_dtype), axis=[k] ), name="C", ) return (query, value, out) def conv2d_winograd_nhwc( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 0, dilation: int = 1, tile_size: int = 4, ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: from tvm.topi.nn.conv2d import ( # pylint: disable=import-outside-toplevel _conv2d_winograd_nhwc_impl, ) target = Target.current(allow_none=True) if target is not None and target.kind.name == "cuda": write_cache_level = 3 else: write_cache_level = 2 data = te.placeholder((N, H, W, CI), "float32", name="data") weight = te.placeholder((kernel_size, kernel_size, CO, CI), "float32", name="weight") out = _conv2d_winograd_nhwc_impl( data, weight, stride, padding, dilation, "float32", pre_computed=True, auto_scheduler_rewritten_layout="", meta_schedule_original_shape=None, tile_size=tile_size, write_cache_level=write_cache_level, ) return (data, weight, out) def conv2d_winograd_nchw( # pylint: disable=invalid-name,missing-docstring N: int, H: int, W: int, CI: int, CO: int, kernel_size: int, stride: int = 1, padding: int = 1, dilation: int = 1, ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: from tvm.topi.cuda.conv2d_winograd import ( # pylint: disable=import-outside-toplevel _infer_tile_size, ) from tvm.topi.nn.conv2d import ( # pylint: disable=import-outside-toplevel _conv2d_winograd_nchw_impl, ) data = te.placeholder((N, CI, H, W), "float32", name="data") weight = te.placeholder((kernel_size, kernel_size, CI, CO), "float32", name="weight") out = _conv2d_winograd_nchw_impl( data, weight, stride, padding, dilation, "float32", pre_computed=True, auto_scheduler_rewritten_layout="", meta_schedule_original_shape=None, tile_size=_infer_tile_size(data, weight), ) return (data, weight, out) def matmul( n: int, m: int, k: int, in_dtype: str = "float32", out_dtype: str = "float32" ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: a = te.placeholder((n, k), name="A", dtype=in_dtype) b = te.placeholder((k, m), name="B", dtype=in_dtype) k = te.reduce_axis((0, k), name="k") c = te.compute( (n, m), lambda i, j: te.sum(a[i, k].astype(out_dtype) * b[k, j].astype(out_dtype), axis=[k]), name="C", ) return (a, b, c) def matmul_relu( n: int, m: int, k: int, in_dtype: str = "float32", out_dtype: str = "float32" ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: a = te.placeholder((n, k), name="A", dtype=in_dtype) b = te.placeholder((k, m), name="B", dtype=in_dtype) k = te.reduce_axis((0, k), name="k") c = te.compute( (n, m), lambda i, j: te.sum(a[i, k].astype(out_dtype) * b[k, j].astype(out_dtype), axis=[k]), name="C", ) d = topi.nn.relu(c) # pylint: disable=invalid-name return (a, b, d) def conv2d_nchw( # pylint: disable=invalid-name n: int, h: int, w: int, ci: int, co: int, kh: int, kw: int, stride: int, padding: int, dilation: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor]: x = te.placeholder((n, ci, h, w), name="X", dtype=in_dtype) w = te.placeholder((co, ci, kh, kw), name="W", dtype=in_dtype) y = topi.nn.conv2d_nchw( Input=x, Filter=w, stride=stride, padding=padding, dilation=dilation, out_dtype=out_dtype ) return (x, w, y) def conv2d_nchw_bias_bn_relu( # pylint: disable=invalid-name n: int, h: int, w: int, ci: int, co: int, kh: int, kw: int, stride: int, padding: int, dilation: int = 1, in_dtype: str = "float32", out_dtype: str = "float32", ) -> Tuple[te.Tensor, te.Tensor, te.Tensor, te.Tensor, te.Tensor, te.Tensor]: oh = (h + 2 * padding - (kh - 1) * dilation - 1) // stride + 1 # pylint: disable=invalid-name ow = (w + 2 * padding - (kw - 1) * dilation - 1) // stride + 1 # pylint: disable=invalid-name x = te.placeholder((n, ci, h, w), name="X", dtype=in_dtype) w = te.placeholder((co, ci, kh, kw), name="W", dtype=in_dtype) b = te.placeholder((co, 1, 1), name="B", dtype=out_dtype) bn_scale = te.placeholder((co, 1, 1), name="bn_scale", dtype=out_dtype) bn_offset = te.placeholder((co, 1, 1), name="bn_offset", dtype=out_dtype) y = topi.nn.conv2d_nchw( Input=x, Filter=w, stride=stride, padding=padding, dilation=dilation, out_dtype=out_dtype ) y = te.compute((n, co, oh, ow), lambda i, j, k, l: y[i, j, k, l] + b[j, 0, 0], name="bias_add") y = te.compute( (n, co, oh, ow), lambda i, j, k, l: y[i, j, k, l] * bn_scale[j, 0, 0], name="bn_mul" ) y = te.compute( (n, co, oh, ow), lambda i, j, k, l: y[i, j, k, l] + bn_offset[j, 0, 0], name="bn_add" ) y = topi.nn.relu(y) return (x, w, b, bn_scale, bn_offset, y) def max_pool2d_nchw( # pylint: disable=invalid-name n: int, h: int, w: int, ci: int, padding: int, ) -> Tuple[te.Tensor, te.Tensor]: # pylint: disable=invalid-name x = te.placeholder((n, ci, h, w), name="X") y = topi.nn.pool2d(x, [2, 2], [1, 1], [1, 1], [padding, padding, padding, padding], "max") return (x, y) def softmax_mn(m, n) -> Tuple[te.Tensor, te.Tensor]: # pylint: disable=invalid-name a = te.placeholder((m, n), name="A") b = topi.nn.softmax(a, axis=1) return (a, b) def create_te_workload(name: str, idx: int) -> tir.PrimFunc: workload_func, params = CONFIGS[name] return te.create_prim_func(workload_func(*params[idx])) # type: ignore CONFIGS = { "C1D": ( conv1d_nlc, [ # derived from conv2d_shapes (1, 256, 64, 128, 3, 2, 1), # (1, 256, 64, 128, 1, 2, 0), # (1, 256, 64, 64, 1, 1, 0), # (1, 128, 128, 256, 3, 2, 1), (1, 128, 128, 256, 1, 2, 0), # (1, 128, 128, 128, 3, 1, 1), # (1, 64, 256, 512, 3, 2, 1), # (1, 64, 256, 512, 1, 2, 0), (1, 64, 256, 256, 5, 1, 2), (1, 32, 512, 512, 3, 1, 1), ], ), "C2D": ( conv2d_nhwc, [ # all conv2d layers in resnet-18 (1, 224, 224, 3, 64, 7, 2, 3), # (1, 56, 56, 64, 128, 3, 2, 1), # (1, 56, 56, 64, 128, 1, 2, 0), # (1, 56, 56, 64, 64, 3, 1, 1), (1, 56, 56, 64, 64, 1, 1, 0), # (1, 28, 28, 128, 256, 3, 2, 1), # (1, 28, 28, 128, 256, 1, 2, 0), # (1, 28, 28, 128, 128, 3, 1, 1), # (1, 14, 14, 256, 512, 3, 2, 1), # (1, 14, 14, 256, 512, 1, 2, 0), (1, 14, 14, 256, 256, 3, 1, 1), (1, 7, 7, 512, 512, 3, 1, 1), ], ), "C3D": ( conv3d_ndhwc, [ # Derived from conv2d_shapes. Use depth=16 for all configurations (1, 16, 224, 224, 3, 64, 7, 2, 3), # (1, 16, 56, 56, 64, 128, 3, 2, 1), # (1, 16, 56, 56, 64, 128, 1, 2, 0), # (1, 16, 56, 56, 64, 64, 3, 1, 1), (1, 16, 56, 56, 64, 64, 1, 1, 0), # (1, 16, 28, 28, 128, 256, 3, 2, 1), # (1, 16, 28, 28, 128, 256, 1, 2, 0), # (1, 16, 28, 28, 128, 128, 3, 1, 1), # (1, 16, 14, 14, 256, 512, 3, 2, 1), # (1, 16, 14, 14, 256, 512, 1, 2, 0), (1, 16, 14, 14, 256, 256, 3, 1, 1), (1, 16, 7, 7, 512, 512, 3, 1, 1), ], ), "GMM": ( batch_matmul_nkkm, [ (1, 128, 128, 128), (1, 512, 32, 512), (1, 512, 512, 512), (1, 1024, 1024, 1024), ], ), "GRP": ( conv2d_nhwc, [ # Derived from conv2d_shapes. Use group=4 for all configurations (1, 56, 56, 64, 128, 3, 2, 1, 1, 4), # (1, 56, 56, 64, 128, 1, 2, 0 , 1, 4), # (1, 56, 56, 64, 64, 3, 1, 1 , 1, 4), (1, 56, 56, 64, 64, 1, 1, 0, 1, 4), # (1, 28, 28, 128, 256, 3, 2, 1, 1, 4), # (1, 28, 28, 128, 256, 1, 2, 0, 1, 4), # (1, 28, 28, 128, 128, 3, 1, 1, 1, 4), # (1, 14, 14, 256, 512, 3, 2, 1, 1, 4), # (1, 14, 14, 256, 512, 1, 2, 0, 1, 4), (1, 14, 14, 256, 256, 3, 1, 1, 1, 4), (1, 7, 7, 512, 512, 3, 1, 1, 1, 4), ], ), "DIL": ( conv2d_nhwc, [ # Derived from conv2d_shapes. Use dilation=2 for all configurations (1, 224, 224, 3, 64, 7, 2, 3, 2), # (1, 56, 56, 64, 128, 3, 2, 1 , 2), # (1, 56, 56, 64, 128, 1, 2, 0 , 2), # (1, 56, 56, 64, 64, 3, 1, 1 , 2), (1, 56, 56, 64, 64, 1, 1, 0, 2), # (1, 28, 28, 128, 256, 3, 2, 1, 2), # (1, 28, 28, 128, 256, 1, 2, 0, 2), # (1, 28, 28, 128, 128, 3, 1, 1, 2), # (1, 14, 14, 256, 512, 3, 2, 1, 2), # (1, 14, 14, 256, 512, 1, 2, 0, 2), (1, 14, 14, 256, 256, 3, 1, 1, 2), (1, 7, 7, 512, 512, 3, 1, 1, 2), ], ), "DEP": ( depthwise_conv2d_nhwc, [ # all depthwise conv2d layers in mobilenet (1, 112, 112, 32, 3, 1, 1), (1, 112, 112, 64, 3, 2, 1), # (1, 56, 56, 128, 3, 1, 1), # (1, 56, 56, 128, 3, 2, 1), # (1, 28, 28, 256, 3, 1, 1), # (1, 28, 28, 256, 3, 2, 1), # (1, 14, 14, 512, 3, 1, 1), (1, 14, 14, 512, 3, 2, 1), (1, 7, 7, 1024, 3, 1, 1), ], ), "T2D": ( conv2d_transpose_nhwc, [ # all conv2d transpose layers in DCGAN (1, 4, 4, 512, 256, 4, 2, 1), (1, 8, 8, 256, 128, 4, 2, 1), (1, 16, 16, 128, 64, 4, 2, 1), (1, 32, 32, 64, 3, 4, 2, 1), ], ), "CAP": ( conv2d_capsule_nhwijc, [ # all conv2d capsule layers in matrix capsules withemrouting (ICLR 2018) (1, 16, 16, 32, 32, 3, 2, 1), (1, 8, 8, 32, 32, 3, 1, 1), (1, 16, 16, 8, 16, 3, 2, 1), (1, 8, 8, 16, 16, 3, 1, 1), ], ), "NRM": ( norm_bmn, [ (1, 256, 256), (1, 512, 512), (1, 1024, 1024), (1, 4096, 1024), ], ), "SFM": ( softmax_mn, [ (256, 256), (512, 512), (1024, 1024), (2048, 2048), ], ), "CBR": ( conv2d_nhwc_bn_relu, [ (1, 224, 224, 3, 64, 7, 2, 3), (1, 56, 56, 64, 128, 3, 2, 1), (1, 28, 28, 128, 256, 1, 2, 0), (1, 7, 7, 512, 512, 3, 1, 1), ], ), "TBG": ( transpose_batch_matmul, [ (1, 128, 12, 64), (1, 128, 16, 64), (1, 64, 12, 128), (1, 128, 12, 128), ], ), "C2D_WIN_NHWC": ( conv2d_winograd_nhwc, [ (1, 14, 14, 128, 128, 6), ], ), "C2D_WIN_NCHW": ( conv2d_winograd_nchw, [ (1, 56, 56, 64, 64, 6), ], ), }
29,336
32.41344
99
py
tvm
tvm-main/python/tvm/meta_schedule/testing/tune_utils.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Testing utility functions in meta schedule""" from typing import Callable, Optional, Union, List, Dict from statistics import median import json import warnings import numpy as np # type: ignore import tvm from tvm.runtime import NDArray def generate_input_data( input_shape: List[int], input_dtype: str, *, low: Optional[int] = None, high: Optional[int] = None, ) -> np.ndarray: """Generate input date with given shape and data type. Parameters ---------- input_shape : List[int] The shape of the input data. input_dtype : str The data type of the input date. Returns ------- input_data : np.ndarray The generated input data with given shape and data type in numpy ndarray. """ if input_dtype.startswith("float"): return np.random.uniform(size=input_shape).astype(input_dtype) if low is None or high is None: warnings.warn( f"Model input value range for shape {input_shape} of {input_dtype} is not set!" ) range_map = { "uint8": (0, 255), "int8": (-128, 127), "int32": (0, 10000), "int64": (0, 10000), } if input_dtype in range_map: _low, _high = range_map[input_dtype] return np.random.randint( low=_low if low is None else low, high=_high if high is None else high, size=input_shape, dtype=input_dtype, ) raise ValueError("Unsupported input datatype!") def create_timer(backend: str) -> Callable: """Create a function to run and benchmark the performance of whole given runtime module, or Executable in relay vm. Parameters ---------- backend : str The backend to use, graph / vm. Returns ------- func : Callable The function to benchmark the workload. """ def f_timer( rt_mod: Union[tvm.runtime.Module, tvm.runtime.vm.Executable], dev: tvm.runtime.Device, input_data: Dict[str, NDArray], ) -> None: """Run and benchmark the given runtime module, print out the result. Parameters ---------- rt_mod : Union[tvm.runtime.Module, tvm.runtime.vm.Executable] The runtime module or vm executable. dev : tvm.runtime.Device The device type to run workload. input_data : Dict[str, np.ndarray] The input data as a dictionary. """ from tvm.contrib.graph_executor import GraphModule # pylint:disable=import-outside-toplevel from tvm.runtime.vm import VirtualMachine # pylint:disable=import-outside-toplevel try: if backend == "vm": vm = VirtualMachine(rt_mod, dev) # pylint: disable=invalid-name ftimer = vm.benchmark( dev, min_repeat_ms=500, repeat=5, number=1, end_to_end=False, **input_data ) elif backend == "graph": mod = GraphModule(rt_mod["default"](dev)) for input_name, input_value in input_data.items(): mod.set_input(input_name, input_value) ftimer = mod.module.time_evaluator( "run", dev, min_repeat_ms=500, repeat=5, number=1 )() else: raise ValueError(f"Backend {backend} not supported in f_timer!") results = list(np.array(ftimer.results) * 1000.0) # type: ignore print("Running time in time_evaluator: ", results) print("-------------------------------") print(f" Min (ms) : {min(results)}") print(f" Max (ms) : {max(results)}") print(f" Median (ms) : {median(results)}") print(f"Average (ms) : {sum(results) / len(results)}") except Exception as exc: # pylint: disable=broad-except print( f"Run module f_timer via RPC failed, exception: {exc}", ) return f_timer def create_time_per_layer(graph: str) -> Callable: """Create a function to run and benchmark the per-layer performance of given runtime module, given the graph output of the module from graph compiler. Parameters ---------- graph : str The json format graph output of the module from graph compiler. Returns ------- func : Callable The function using the json format graph. """ def f_time_per_layer( rt_mod: tvm.runtime.Module, dev: tvm.runtime.Device, input_data: Dict[str, NDArray], ) -> None: """Run and benchmark the per-layer performance of given runtime module, print out the result. Parameters ---------- rt_mod : tvm.runtime.Module The runtime module. dev : tvm.runtime.Device The device type to run workload. input_data : Dict[str, np.ndarray] The input data as a dictionary. """ # pylint:disable=import-outside-toplevel from tvm.contrib.debugger.debug_executor import create # pylint:enable=import-outside-toplevel try: mod = create(graph, rt_mod, dev) for input_name, input_value in input_data.items(): mod.set_input(input_name, input_value) graph_nodes = [n["name"] for n in json.loads(graph)["nodes"]] graph_time = mod.run_individual(number=10, repeat=1, min_repeat_ms=5000) print("Running time of each layer:") print("---------------------------") print("|graph_nodes| = ", len(graph_nodes)) print("|graph_time| = ", len(graph_time)) for k, v in zip(graph_nodes, graph_time): print(k, float(v) * 1e6, "us") except Exception as exc: # pylint: disable=broad-except print( f"Run module f_time_per_layer via RPC failed, exception: {exc}", ) return f_time_per_layer def create_calculator(backend: str) -> Callable: """Create a function to fetch the computing result of running the given runtime module. Parameters ---------- backend : str The backend to use, only tir is supported for now. Returns ------- func : Callable The function to fetch the computing result. """ def f_calculator( rt_mod: tvm.runtime.Module, dev: tvm.runtime.Device, # pylint: disable=unused-argument input_data: Dict[str, NDArray], ) -> List[NDArray]: """Fetch the result of running the given runtime module. Parameters ---------- rt_mod : Union[tvm.runtime.Module, tvm.runtime.vm.Executable] The runtime module or vm executable. dev : tvm.device The device type to run workload. input_data : Dict[str, np.ndarray] The input data as a dictionary. """ try: if backend == "tir": data = [v for _, v in sorted(input_data.items(), key=lambda x: x[0])] rt_mod(*data) return data else: raise ValueError(f"Backend {backend} not supported in f_calculator!") except Exception as exc: # pylint: disable=broad-except print( f"Run module f_calculator via RPC failed, exception: {exc}", ) return None return f_calculator
8,241
33.057851
100
py
tvm
tvm-main/python/tvm/meta_schedule/testing/dataset_collect_models.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import os from typing import List, Tuple from tqdm import tqdm # type: ignore from tvm.meta_schedule.testing.relay_workload import get_network # pylint: disable=too-many-branches def _build_dataset() -> List[Tuple[str, List[int]]]: network_keys = [] for name in [ "resnet_18", "resnet_50", "mobilenet_v2", "mobilenet_v3", "wide_resnet_50", "resnext_50", "densenet_121", "vgg_16", ]: for batch_size in [1, 4, 8]: for image_size in [224, 240, 256]: network_keys.append((name, [batch_size, 3, image_size, image_size])) # inception-v3 for name in ["inception_v3"]: for batch_size in [1, 2, 4]: for image_size in [299]: network_keys.append((name, [batch_size, 3, image_size, image_size])) # resnet3d for name in ["resnet3d_18"]: for batch_size in [1, 2, 4]: for image_size in [112, 128, 144]: network_keys.append((name, [batch_size, 3, image_size, image_size, 16])) # bert for name in ["bert_tiny", "bert_base", "bert_medium", "bert_large"]: for batch_size in [1, 2, 4]: for seq_length in [64, 128, 256]: network_keys.append((name, [batch_size, seq_length])) # dcgan for name in ["dcgan"]: for batch_size in [1, 4, 8]: for image_size in [64]: network_keys.append((name, [batch_size, 3, image_size, image_size])) return network_keys def main(): model_cache_dir = args.model_cache_dir try: os.makedirs(model_cache_dir, exist_ok=True) except OSError: print(f"Directory {model_cache_dir} cannot be created successfully.") keys = _build_dataset() for name, input_shape in tqdm(keys): get_network(name=name, input_shape=input_shape, cache_dir=model_cache_dir) if __name__ == "__main__": parser = argparse.ArgumentParser() # pylint: disable=invalid-name parser.add_argument( "--model_cache_dir", type=str, help="Please provide the full path to the model cache dir.", ) args = parser.parse_args() # pylint: disable=invalid-name main()
3,050
34.476744
88
py
tvm
tvm-main/python/tvm/meta_schedule/testing/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Testing utilities in meta schedule""" # NOTE: Do not import any module here by default
876
42.85
62
py
tvm
tvm-main/python/tvm/meta_schedule/testing/local_rpc.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """RPC tracker and server running locally""" from tvm.rpc.tracker import Tracker from tvm.rpc.server import Server class LocalRPC: """A pair of RPC tracker/server running locally Parameters ---------- tracker_host : str The host URL of the tracker tracker_port : int The port of the tracker tracker_key: str The key used in the tracker to refer to a worker """ tracker_host: str tracker_port: int tracker_key: str def __init__( self, tracker_key: str = "key", silent: bool = False, no_fork: bool = False, ) -> None: self.tracker = Tracker( silent=silent, port=9190, port_end=12345, ) self.server = Server( host="0.0.0.0", is_proxy=False, tracker_addr=(self.tracker.host, self.tracker.port), key=tracker_key, silent=silent, no_fork=no_fork, port=9190, port_end=12345, ) self.tracker_host = self.tracker.host self.tracker_port = self.tracker.port self.tracker_key = tracker_key def __enter__(self): return self def __exit__(self, _type, _value, _traceback): if hasattr(self, "server"): del self.server if hasattr(self, "tracker"): del self.tracker
2,186
29.375
64
py
tvm
tvm-main/python/tvm/meta_schedule/testing/space_generation.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-module-docstring,missing-function-docstring,missing-class-docstring from typing import List, Optional, Tuple, Union # isort: off from typing_extensions import Literal # isort: on from tvm import meta_schedule as ms from tvm.ir import IRModule, structural_equal from tvm.target import Target from tvm.tir import Schedule from tvm.tir.schedule import Trace from tvm.tir.schedule.testing import verify_trace_roundtrip def get_rules( kind: Literal["llvm", "cuda", "cuda-tensorcore", "hexagon"], types: Union[type, Tuple[type, ...]], ) -> List[ms.ScheduleRule]: """Get default schedule rules""" rules = ms.ScheduleRule.create(kind) return [rule for rule in rules if isinstance(rule, types)] def generate_design_space( kind: Literal["llvm", "cuda", "cuda-tensorcore", "hexagon"], mod: IRModule, target: Target, types: Union[type, Tuple[type, ...]], sch_rules: Optional[List[ms.ScheduleRule]] = None, ) -> List[Schedule]: if sch_rules is None: sch_rules = get_rules(kind, types) else: assert types is None return ms.TuneContext( mod=mod, target=target, space_generator=ms.space_generator.PostOrderApply( sch_rules=sch_rules, postprocs=[], mutator_probs={}, ), task_name="test", ).generate_design_space() def _find_match_sketch_id( mod: IRModule, sketches: List[Schedule], expected_mod: IRModule, expected_decision: List[Tuple[str, List[int]]], *, debug_mask="all", ) -> Optional[int]: for sketch_id, sketch in enumerate(sketches): i = 0 new_decisions = {} for inst in sketch.trace.insts: if not inst.kind.name.startswith("Sample"): continue assert i < len(expected_decision) if inst.kind.name == expected_decision[i][0]: new_decisions[inst] = expected_decision[i][1] i += 1 if len(new_decisions) != len(expected_decision): continue sch = Schedule(mod, debug_mask=debug_mask) Trace( insts=sketch.trace.insts, decisions=new_decisions, ).apply_to_schedule(sch, remove_postproc=True) if structural_equal(sch.mod, expected_mod): verify_trace_roundtrip(sch=sch, mod=mod, debug_mask=debug_mask, text_format="json") return sketch_id return None def check_sketches( mod: IRModule, sketches: List[Schedule], expected_mods: List[IRModule], expected_decisions: List[List[Tuple[str, List[int]]]], *, debug_mask="all", ): assert len(expected_mods) == len(expected_decisions) assert len(sketches) == len(expected_mods) expected_mods = [ IRModule({"main": m}) if not isinstance(m, IRModule) else m for m in expected_mods ] sketches = list(sketches) for expected_id, (expected_mod, expected_decision) in enumerate( zip(expected_mods, expected_decisions) ): sketch_id = _find_match_sketch_id( mod, sketches, expected_mod, expected_decision, debug_mask=debug_mask, ) if sketch_id is None: raise AssertionError( f"Expected sketch #{expected_id} doesn't exist in the generated sketches." ) sketches.pop(sketch_id) def print_sketches(sketches: List[Schedule]): for i, sch in enumerate(sketches): print(f"###### {i}") sch.mod.show(black_format=False) for inst in sch.trace.insts: if inst in sch.trace.decisions: print(f'("{inst.kind.name}", {sch.trace.decisions[inst]}),')
4,511
32.671642
95
py
tvm
tvm-main/python/tvm/meta_schedule/testing/relay_workload.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Workloads in Relay IR""" # pylint: disable=import-outside-toplevel import logging import multiprocessing import os import pickle from typing import Any, Dict, List, Optional, Tuple import tvm import tvm.relay.testing from tvm import meta_schedule as ms from tvm import relay from tvm.ir import IRModule from tvm.runtime import NDArray, load_param_dict, save_param_dict from tvm.target import Target logger = logging.getLogger(__name__) # pylint: disable=invalid-name def _get_network( args: Tuple[str, List[int], Optional[str]] ) -> Tuple[IRModule, bytearray, Tuple[str, List[int], str]]: name: str input_shape: List[int] layout: Optional[str] name, input_shape, layout = args if layout == "None": layout = None mod: IRModule if name in [ "resnet_18", "resnet_50", "wide_resnet_50", "resnext_50", "mobilenet_v2", "mobilenet_v3", "inception_v3", "densenet_121", "resnet3d_18", "vgg_16", ]: import torch # type: ignore from torchvision import models # type: ignore assert layout is None or layout in ["NCHW", "NHWC"] params: Dict[str, Any] = {} if name in ["resnet_18", "resnet_50"]: model = getattr(models, name.replace("_", "")) elif name == "wide_resnet_50": model = getattr(models, "wide_resnet50_2") elif name == "resnext_50": model = getattr(models, "resnext50_32x4d") elif name == "mobilenet_v2": model = getattr(models, name) elif name == "mobilenet_v3": model = getattr(models, name + "_large") elif name == "inception_v3": model = getattr(models, name) params["aux_logits"] = False elif name == "densenet_121": model = getattr(models, name.replace("_", "")) elif name == "resnet3d_18": model = models.video.r3d_18 elif name == "vgg_16": model = getattr(models, name.replace("_", "")) try: model = model(**params, weights=None) except TypeError: model = model(**params, pretrained=False) dtype = "float32" input_data = torch.randn(input_shape).type( # pylint: disable=no-member { "float32": torch.float32, # pylint: disable=no-member }[dtype] ) scripted_model = torch.jit.trace(model, input_data).eval() # type: ignore input_name = "input0" shape_list = [(input_name, input_shape)] mod, params = relay.frontend.from_pytorch(scripted_model, shape_list) passes = [relay.transform.RemoveUnusedFunctions()] if layout is None or layout == "NHWC": # PyTorch is imported as NCHW by default passes.append( relay.transform.ConvertLayout( { "nn.conv2d": ["NHWC", "default"], "nn.conv3d": ["NDHWC", "default"], "nn.max_pool2d": ["NHWC", "default"], "nn.avg_pool2d": ["NHWC", "default"], } ) ) with tvm.transform.PassContext(opt_level=3): mod = tvm.transform.Sequential(passes)(mod) inputs = (input_name, input_shape, dtype) elif name in ["bert_tiny", "bert_base", "bert_medium", "bert_large"]: os.environ["TOKENIZERS_PARALLELISM"] = "false" # pip3 install transformers==3.5 torch==1.7 import torch # type: ignore import transformers # type: ignore assert layout is None config_dict = { "bert_tiny": transformers.BertConfig( num_hidden_layers=6, hidden_size=512, intermediate_size=2048, num_attention_heads=8, return_dict=False, ), "bert_base": transformers.BertConfig( num_hidden_layers=12, hidden_size=768, intermediate_size=3072, num_attention_heads=12, return_dict=False, ), "bert_medium": transformers.BertConfig( num_hidden_layers=12, hidden_size=1024, intermediate_size=4096, num_attention_heads=16, return_dict=False, ), "bert_large": transformers.BertConfig( num_hidden_layers=24, hidden_size=1024, intermediate_size=4096, num_attention_heads=16, return_dict=False, ), } configuration = config_dict[name] model = transformers.BertModel(configuration) input_name = "input_ids" input_dtype = "int64" a = torch.randint(10000, input_shape) # pylint: disable=no-member model.eval() scripted_model = torch.jit.trace(model, [a], strict=False) # type: ignore input_name = "input_ids" shape_list = [(input_name, input_shape)] mod, params = relay.frontend.from_pytorch(scripted_model, shape_list) mod = relay.transform.FastMath()(mod) mod = relay.transform.CombineParallelBatchMatmul()(mod) inputs = (input_name, input_shape, input_dtype) elif name == "dcgan": assert layout is None output_shape = input_shape batch_size = output_shape[0] oshape = output_shape[1:] mod, params = relay.testing.dcgan.get_workload( batch_size=batch_size, oshape=oshape, layout="NHWC", ) inputs = ("data", [100], "float32") else: raise ValueError("Invalid name: " + name) params_bytearray: bytearray = save_param_dict(params) return mod, params_bytearray, inputs def _load_cache(cache_dir: Optional[str], filename: str) -> Optional[List[Any]]: if cache_dir is None: return None path = os.path.join(os.path.expanduser(cache_dir), filename) if not os.path.exists(path): return None logger.info("Loaded from cached: %s", path) with open(path, "rb") as i_f: return pickle.load(i_f) def _save_cache(cache_dir: Optional[str], filename: str, objects: List[Any]) -> None: if cache_dir is None: return path = os.path.join(os.path.expanduser(cache_dir), filename) with open(path, "wb") as o_f: pickle.dump(objects, o_f) def get_network( name: str, input_shape: List[int], *, layout: Optional[str] = None, cache_dir: Optional[str] = None, ) -> Tuple[IRModule, Dict[str, NDArray], Tuple[str, List[int], str]]: """Get the symbol definition and random weight of a network Parameters ---------- name : str The name of the network. input_shape : List[int] The shape of the input tensor. layout : Optional[str] The layout of the input tensor. For vision models, the layout is by default NHWC. cache_dir : Optional[str], optional The directory to cache the generated network. If not specified, the cache will be disabled. Returns ------- mod : IRModule The IRModule representing the network. params : Dict[str, NDArray] The parameters of the networks. inputs : Tuple[str, List[int], str] The name, shape and dtype of the input tensor. """ mod: IRModule params: Dict[str, NDArray] inputs: Tuple[str, List[int], str] params_bytearray: bytearray filename = f'relay-{name}-{layout}-{",".join(str(i) for i in input_shape)}.json' cached = _load_cache(cache_dir, filename) if cached is None: with multiprocessing.Pool(processes=1) as pool: result = pool.map(_get_network, [(name, input_shape, layout)]) ((mod, params_bytearray, inputs),) = result cached = [mod, params_bytearray, inputs] _save_cache(cache_dir, filename, cached) mod, params_bytearray, inputs = cached params = load_param_dict(params_bytearray) return mod, params, inputs def extract_from_relay( mod: IRModule, target: Target, params: Optional[Dict[str, NDArray]], name: str, input_shape: List[int], *, cache_dir: Optional[str] = None, ) -> List[ms.ExtractedTask]: """Extract the tasks from a network. Parameters ---------- mod : IRModule The IRModule representing the network. target : Target The target that the network will be deployed to. params : Optional[Dict[str, NDArray]] The parameters of the networks. name : str The name of the network. input_shape : List[int] The shape of the input tensor. cache_dir : Optional[str] The directory to cache the generated network. If not specified, the cache will be disabled. Returns ------- extracted_tasks : List[ExtractedTask] The extracted tasks. """ filename = f'tasks-{target.kind.name}-{name}-{",".join(str(i) for i in input_shape)}.json' extracted_tasks = _load_cache(cache_dir, filename) if extracted_tasks is None: extracted_tasks = ms.relay_integration.extract_tasks( mod=mod, target=target, params=params, ) extracted_tasks = list(extracted_tasks) _save_cache(cache_dir, filename, extracted_tasks) return extracted_tasks SUPPORTED = [ # TorchVision "resnet_18", "resnet_50", "mobilenet_v2", "mobilenet_v3", "wide_resnet_50", "resnext_50", "resnet3d_18", "inception_v3", "densenet_121", "vgg_16", # Transformer "bert_tiny", "bert_base", "bert_medium", "bert_large", # Relay testing "dcgan", ]
10,611
32.371069
94
py
tvm
tvm-main/python/tvm/meta_schedule/testing/distributed_measure_candidates.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import glob import os from tqdm import tqdm # type: ignore from tvm import meta_schedule as ms from tvm.target import Target def _parse_args(): parser = argparse.ArgumentParser() parser.add_argument( "--candidate_cache_dir", type=str, help="Please provide the full path to the candidates." ) parser.add_argument( "--result_cache_dir", type=str, help="Please provide the full path to the result database." ) parser.add_argument( "--target", type=str, default="nvidia/nvidia-v100", help="Please specify the target hardware for tuning context.", ) parser.add_argument( "--rpc_host", type=str, help="Please provide the private IPv4 address for the tracker." ) parser.add_argument( "--rpc_port", type=int, default=4445, help="Please provide the port for the tracker." ) parser.add_argument( "--rpc_key", type=str, default="p3.2xlarge", help="Please provide the key for the rpc servers.", ) parser.add_argument( "--builder_timeout_sec", type=int, default=10, help="The time for the builder session to time out.", ) parser.add_argument( "--min_repeat_ms", type=int, default=100, help="The time for preheating the gpu." ) parser.add_argument( "--runner_timeout_sec", type=int, default=100, help="The time for the runner session to time out.", ) parser.add_argument( "--cpu_flush", type=bool, default=False, help="Whether to enable cpu cache flush or not." ) parser.add_argument( "--batch_size", type=int, default=128, help="The batch size of candidates sent to builder and runner each time.", ) return parser.parse_args() # pylint: disable=too-many-locals def measure_candidates(database, builder, runner): """Send the candidates to builder and runner for distributed measurement, and save the results in a new json database. Parameters ---------- database : JSONDatabase The database for candidates to be measured. builder : Builder The builder for building the candidates. runner : Runner The runner for measuring the candidates. Returns ------- None """ candidates, runner_results, build_fail_indices, run_fail_indices = [], [], [], [] context = ms.TuneContext(target=Target(args.target)) tuning_records = database.get_all_tuning_records() for record in tuning_records: candidates.append(record.as_measure_candidate()) with ms.Profiler() as profiler: for idx in range(0, len(candidates), args.batch_size): batch_candidates = candidates[idx : idx + args.batch_size] context._set_measure_candidates(batch_candidates) # pylint: disable=protected-access with ms.Profiler.timeit("build"): context._send_to_builder(builder) # pylint: disable=protected-access with ms.Profiler.timeit("run"): context._send_to_runner(runner) # pylint: disable=protected-access batch_runner_results = context._join() # pylint: disable=protected-access runner_results.extend(batch_runner_results) for i, result in enumerate(context.builder_results): if result.error_msg is None: ms.utils.remove_build_dir(result.artifact_path) else: build_fail_indices.append(i + idx) context._clear_measure_state() # pylint: disable=protected-access model_name, workload_name = database.path_workload.split("/")[-2:] record_name = database.path_tuning_record.split("/")[-1] new_database = ms.database.JSONDatabase( path_workload=os.path.join(args.result_cache_dir, model_name, workload_name), path_tuning_record=os.path.join(args.result_cache_dir, model_name, record_name), ) workload = tuning_records[0].workload new_database.commit_workload(workload.mod) for i, (record, result) in enumerate(zip(tuning_records, runner_results)): if result.error_msg is None: new_database.commit_tuning_record( ms.database.TuningRecord( trace=record.trace, workload=workload, run_secs=[v.value for v in result.run_secs], target=Target(args.target), ) ) else: run_fail_indices.append(i) fail_indices_name = workload_name.replace("_workload.json", "_failed_indices.txt") with open( os.path.join(args.result_cache_dir, model_name, fail_indices_name), "w", encoding="utf8" ) as file: file.write(" ".join([str(n) for n in run_fail_indices])) print( f"Builder time: {profiler.get()['build']}, Runner time: {profiler.get()['run']}\n\ Failed number of builds: {len(build_fail_indices)},\ Failed number of runs: {len(run_fail_indices)}" ) args = _parse_args() # pylint: disable=invalid-name def main(): builder = ms.builder.LocalBuilder(timeout_sec=args.builder_timeout_sec) runner = ms.runner.RPCRunner( rpc_config=ms.runner.RPCConfig( tracker_host=args.rpc_host, tracker_port=args.rpc_port, tracker_key=args.rpc_key, session_timeout_sec=args.runner_timeout_sec, ), evaluator_config=ms.runner.EvaluatorConfig( number=3, repeat=1, min_repeat_ms=args.min_repeat_ms, enable_cpu_cache_flush=args.cpu_flush, ), max_workers=os.cpu_count(), ) if not os.path.isdir(args.candidate_cache_dir): raise Exception("Please provide a correct candidate cache dir.") try: os.makedirs(args.result_cache_dir, exist_ok=True) except OSError: print(f"Directory {args.result_cache_dir} cannot be created successfully.") model_dirs = glob.glob(os.path.join(args.candidate_cache_dir, "*")) for model_dir in model_dirs: model_name = model_dir.split("/")[-1] os.makedirs(os.path.join(args.result_cache_dir, model_name), exist_ok=True) all_tasks = glob.glob(os.path.join(model_dir, "*.json")) workload_paths = [] for path in all_tasks: if path.endswith("_workload.json"): workload_paths.append(path) for workload_path in tqdm(workload_paths): candidate_path = workload_path.replace("_workload.json", "_candidates.json") database = ms.database.JSONDatabase( path_workload=workload_path, path_tuning_record=candidate_path, ) measure_candidates(database, builder, runner) if __name__ == "__main__": main()
7,702
37.708543
99
py
tvm
tvm-main/python/tvm/meta_schedule/testing/tune_relay.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=missing-docstring import argparse import json import logging from distutils.util import strtobool from typing import Dict import numpy as np # type: ignore import tvm from tvm import meta_schedule as ms from tvm.meta_schedule.testing.custom_builder_runner import run_module_via_rpc from tvm.meta_schedule.testing.relay_workload import get_network from tvm.meta_schedule.testing.tune_utils import create_timer, generate_input_data from tvm.support import describe def _parse_args(): args = argparse.ArgumentParser() args.add_argument( "--workload", type=str, required=True, ) args.add_argument( "--input-shape", type=str, required=True, ) args.add_argument( "--target", type=str, required=True, ) args.add_argument( "--num-trials", type=int, required=True, ) args.add_argument( "--rpc-host", type=str, required=True, ) args.add_argument( "--rpc-port", type=int, required=True, ) args.add_argument( "--rpc-key", type=str, required=True, ) args.add_argument( "--work-dir", type=str, required=True, ) args.add_argument( "--layout", type=str, default=None, ) args.add_argument( "--cache-dir", type=str, default=None, ) args.add_argument( "--number", type=int, default=3, ) args.add_argument( "--repeat", type=int, default=1, ) args.add_argument( "--min-repeat-ms", type=int, default=100, ) args.add_argument( "--adaptive-training", type=lambda x: bool(strtobool(x)), help="example: True / False", default=True, ) args.add_argument( "--cpu-flush", type=lambda x: bool(strtobool(x)), help="example: True / False", required=True, ) args.add_argument( "--backend", type=str, choices=["graph", "vm"], help="example: graph / vm", required=True, ) parsed = args.parse_args() parsed.target = tvm.target.Target(parsed.target) parsed.input_shape = json.loads(parsed.input_shape) parsed.rpc_config = ms.runner.RPCConfig( tracker_host=parsed.rpc_host, tracker_port=parsed.rpc_port, tracker_key=parsed.rpc_key, session_timeout_sec=600, ) return parsed logging.basicConfig( format="%(asctime)s.%(msecs)03d %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S" ) logging.getLogger("tvm.meta_schedule").setLevel(logging.DEBUG) ARGS = _parse_args() def main(): describe() print(f"Workload: {ARGS.workload}") mod, params, (input_name, input_shape, input_dtype) = get_network( ARGS.workload, ARGS.input_shape, layout=ARGS.layout, cache_dir=ARGS.cache_dir, ) input_info = [ { "name": input_name, "shape": input_shape, "dtype": input_dtype, }, ] input_data: Dict[str, np.ndarray] = { item["name"]: generate_input_data( # type: ignore item["shape"], # type: ignore item["dtype"], # type: ignore ) for item in input_info } for item in input_info: print(f" input_name : {item['name']}") print(f" input_shape: {item['shape']}") print(f" input_dtype: {item['dtype']}") with ms.Profiler() as profiler: database = ms.relay_integration.tune_relay( mod=mod, target=ARGS.target, work_dir=ARGS.work_dir, max_trials_global=ARGS.num_trials, num_trials_per_iter=64, params=params, runner=ms.runner.RPCRunner( # type: ignore rpc_config=ARGS.rpc_config, evaluator_config=ms.runner.EvaluatorConfig( number=ARGS.number, repeat=ARGS.repeat, min_repeat_ms=ARGS.min_repeat_ms, enable_cpu_cache_flush=ARGS.cpu_flush, ), alloc_repeat=1, ), cost_model=ms.cost_model.XGBModel( # type: ignore extractor=ms.feature_extractor.PerStoreFeature(), adaptive_training=ARGS.adaptive_training, ), strategy=ms.search_strategy.EvolutionarySearch(), ) lib = ms.relay_integration.compile_relay( database=database, mod=mod, target=ARGS.target, params=params, backend=ARGS.backend, ) print("Tuning Time:") print(profiler.table()) run_module_via_rpc( rpc_config=ARGS.rpc_config, lib=lib, dev_type=ARGS.target.kind.name, args=input_data, continuation=create_timer(ARGS.backend), backend=ARGS.backend, ) if __name__ == "__main__": main()
5,867
26.420561
91
py
tvm
tvm-main/python/tvm/meta_schedule/testing/torchbench/utils.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name, pointless-exception-statement """ Helper functions for running TorchBench through the benchmark functions from TorchDynamo. """ import functools import os import sys from dataclasses import dataclass from enum import Enum from typing import Set import torch # type: ignore class DisallowedOperator(Enum): """ The operators to disallow in the fx graph produced by TorchDynamo. This is to workaround the limitation in TVM's PyTorch frontend. - inplace_copy: aten::copy_ as inplace assign A[...] = ..., or method call A.copy_(...) - einsum: torch.functional.einsum - multihead_attention: torch.nn.MultiheadAttention - as_stride: Tensor.as_stride """ INPLACE_COPY = "inplace_copy" EINSUM = "einsum" MULTIHEAD_ATTENTION = "multihead_attention" AS_STRIDE = "as_stride" def find_torchdynamo() -> str: """ Find the directory of TorchDynamo repo. It can't directly import the benchmark runner in TorchDynamo becuase it isn't designed to be used as a Python package. """ candidates = [ "torchdynamo", "../torchdynamo", "../../torchdynamo", ] for library_dir in candidates: if os.path.exists(f"{library_dir}/benchmarks"): return library_dir raise RuntimeError( """ Cannot find directory for torchdynamo. You need to clone https://github.com/pytorch/torchdynamo to the parent directory of cwd. """ ) DYNAMO_DIR = find_torchdynamo() sys.path.insert( 0, DYNAMO_DIR ) # opacus_cifar10 depends on opacus, which installs a package called 'benchmarks' sys.path.append(f"{DYNAMO_DIR}/benchmarks") # pylint: disable=wrong-import-position, unused-import import torchdynamo # type: ignore from benchmarks.common import same, timed # type: ignore from torchbench import TorchBenchmarkRunner # type: ignore # pylint: disable=wrong-import-position, unused-import def _disallow_operators(disallowed_ops: Set[DisallowedOperator]): """ Disallow certain operators in the fx graph produced by TorchDynamo. There are two ways to disallow operator in TorchDynamo, 1. Use the disallow_in_graph API, which only applies to free function call. 2. Patch the TensorVariable class, which applies to method call on torch.Tensor. """ disallowed_tensor_methods: Set[str] = set() if DisallowedOperator.INPLACE_COPY in disallowed_ops: torchdynamo.disallow_in_graph(torch.Tensor.copy_) disallowed_tensor_methods.update({"copy_", "__setitem__"}) if DisallowedOperator.EINSUM in disallowed_ops: torchdynamo.disallow_in_graph(torch.functional.einsum) if DisallowedOperator.MULTIHEAD_ATTENTION in disallowed_ops: torchdynamo.disallow_in_graph(torch.nn.MultiheadAttention) if DisallowedOperator.AS_STRIDE in disallowed_ops: disallowed_tensor_methods.add("as_stride") tensor_variable_cls = torchdynamo.variables.tensor.TensorVariable old_call_method = tensor_variable_cls.call_method @functools.wraps(old_call_method) def call_method(self, translator, name, args, kwargs): if name in disallowed_tensor_methods: raise torchdynamo.exc.Unsupported(f"Tensor.{name} not supported by TVM.") return old_call_method(self, translator, name, args, kwargs) tensor_variable_cls.call_method = call_method def load_torchdynamo_benchmark_runner( is_cuda: bool, cosine_similarity: bool = False, float32: bool = False, disallowed_operators: Set[DisallowedOperator] = None, ) -> TorchBenchmarkRunner: """ Load the benchmark runner from TorchDynamo. """ @dataclass class RunnerArgs: """ This class simulates the parsed args required by the benchmark code from TorchDynamo. """ ci: bool = False # Whether runs in CI mode. pylint: disable=invalid-name training: bool = False # Whether it benchmarks training workload. use_eval_mode: bool = True # Whether the model should be in eval mode. dynamic_shapes: bool = False # Whether runs the model in dynamic shape mode. float16: bool = False # Whether to cast model and inputs to float16 float32: bool = False # Whether to cast model and inputs to float32 accuracy: bool = False # Whether to perform a accuracy test performance: bool = True # Whether to perform a performance test cosine: bool = False # Whether to use consine similarity to check if output is correct. args = RunnerArgs(cosine=cosine_similarity, float32=float32) runner = TorchBenchmarkRunner() runner.args = args runner.model_iter_fn = runner.forward_pass if disallowed_operators: _disallow_operators(disallowed_operators) if is_cuda: # pylint: disable=import-outside-toplevel import benchmarks.common # type: ignore # pylint: enable=import-outside-toplevel benchmarks.common.synchronize = torch.cuda.synchronize return runner
5,837
33.75
96
py
tvm
tvm-main/python/tvm/meta_schedule/testing/torchbench/run.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=unused-variable """ This script is for benchmarking TVM performance on models from TorchBench. It uses the TorchDynamo as the frontend to ingest models into TVM, and it also leverages the benchmark util from TorchDynamo. TorchDynamo (https://github.com/pytorch/torchdynamo) and TorchBench (https://github.com/pytorch/benchmark) need to be in the parent directory of TVM. We need a local clone of these repos because torchbench and the benchmark runner in TorchDynamo isn't designed to be used as a Python package. To setup the environment, run the following commands in the parent directory of TVM and with the appropriate Python environment: ```bash # torchdynamo requires nightly pytorch. If it fails to find the specified version, try # installing the latest nightly pytorch. pip3 install --pre \ --extra-index-url https://download.pytorch.org/whl/nightly/cu116 \ torch==1.13.0.dev20220926 \ torchvision==0.14.0.dev20220926 \ torchtext==0.14.0.dev20220926 git clone https://github.com/pytorch/torchdynamo pushd torchdynamo git checkout c537639f9712621dc04ca09908796dbbe86c354b pip install -e . popd sudo apt install git-lfs # git lfs is used for TorchBench git clone https://github.com/pytorch/benchmark pushd benchmark python install.py --continue_on_fail # fambench_xlmr might fail to install popd ``` To run a benchmark, the script can be run under 'tune' mode by ```bash python python/tvm/meta_schedule/testing/torchbench/run.py \ --mode tune \ --model resnet50 \ --target "nvidia/geforce-rtx-3070" \ --work-dir /path/to/work/dir/ \ --num-trials 20000 \ --rpc-host <rpc tracker host for tuning> \ --rpc-port <rpc tracker port for tuning> \ --rpc-key <rpc key> \ ``` All available target tags (like nvidia/geforce-rtx-3070) can be found at https://github.com/apache/tvm/blob/main/src/target/tag.cc Then the script can be run under 'eval' mode to actual benchmark the performance, using the tuning database under the work directory. This can be executed on a different machine than the one executes tuning (the database json files need to be inside of the work directory). ```bash python python/tvm/meta_schedule/testing/torchbench/run.py \ --mode eval \ --model resnet50 \ --target "nvidia/geforce-rtx-3070" \ --work-dir /path/to/work/dir/ \ --num-trials 0 ``` Alternatively, both tuning and evaluation can be done in a single run on the same machine, by ```bash python python/tvm/meta_schedule/testing/torchbench/run.py \ --mode all \ --model resnet50 \ --target "llvm -num-cores 6" \ --work-dir /path/to/work/dir/ \ --num-trials 0 ``` """ # pylint: disable=logging-format-interpolation import argparse import contextlib import logging import os import pickle import sys import warnings from collections import defaultdict from enum import Enum from typing import Callable, Dict, List, Tuple import numpy as np # type: ignore import torch # type: ignore from scipy.stats import ttest_ind # type: ignore import tvm import tvm.relay from tvm import meta_schedule as ms from tvm._ffi import get_global_func from tvm.contrib.graph_executor import GraphModule from tvm.meta_schedule.testing.torchbench.utils import ( DisallowedOperator, load_torchdynamo_benchmark_runner, same, timed, ) from tvm.runtime.vm import VirtualMachine from tvm.support import describe # Needs to be imported after the .utils is executed import torchdynamo # type: ignore # isort: skip, pylint: disable=wrong-import-order class RunMode(Enum): """ The running mode of this script. Available values are: - extract: Only import the model and extract tuning tasks from it. - tune: Only tune the tasks and create the tuning database. - eval: Only benchmark model using pre-existing tuning database. - all: Run both tuning and benchmark """ ALL = "all" EXTRACT = "extract" TUNE = "tune" EVAL = "eval" @property def should_extract(self): """ Returns whether it should extract tuning tasks. """ return self in (RunMode.ALL, RunMode.EXTRACT) @property def should_tune(self): """ Returns whether it should tune the tasks. """ return self in (RunMode.ALL, RunMode.TUNE) @property def should_eval(self): """ Returns whether it should actually benchmark the model. """ return self in (RunMode.ALL, RunMode.EVAL) class ResultComparisonMetric(Enum): """ This changes how it compares the results with the expected value during accuracy check. - cosine: Use the cosine similarity. It should be greater than 0.99. - allclose-1e-4: Use the max elementwise absolute difference. It should be less than 1e-4. """ COSINE = "cosine" ALLCLOSE = "allclose-1e-4" def parse_args(): """ Parse arguments """ args = argparse.ArgumentParser() args.add_argument( "--mode", type=RunMode, default=RunMode.ALL, help=RunMode.__doc__, ) args.add_argument( "--batch-size", type=int, default=None, help="The batch size of model input. Use TorchBench's default value if not specified.", ) args.add_argument( "--result-metric", type=ResultComparisonMetric, default=ResultComparisonMetric.ALLCLOSE, help=ResultComparisonMetric.__doc__, ) args.add_argument( "--benchmark-repeat", type=int, default=10, help="The number of times to repeat the benchmark measurement.", ) args.add_argument( "--benchmark-warmup-rounds", type=int, default=5, help="The number of rounds to warmup before starting to measure the performance.", ) args.add_argument( "--disallowed-op", type=str, default="all", help=DisallowedOperator.__doc__, ) # Model selection args.add_argument( "--model", type=str, required=True, help=""" The name of model to run. It should a directory name under https://github.com/pytorch/benchmark/tree/main/torchbenchmark/models. """, ) args.add_argument( "--float32", action="store_true", help=""" Cast model and inputs to fp32 """, ) # Tuning-related config args.add_argument( "--target", type=tvm.target.Target, required=True, help="The target to tune and run benchmark for.", ) args.add_argument( "--work-dir", type=str, required=True, help=""" The working directory to save intermediate results and store databases for compilation. """, ) args.add_argument( "--strategy", type=str, default="evolutionary", help="The search strategy used by MetaSchdule.", ) args.add_argument( "--num-trials", type=int, required=True, help="The max number of trials to run MetaSchedule.", ) args.add_argument( "--max-trials-per-task", type=int, default=None, help=""" The max number of trials to run per task extracted in MetaSchedule. By default it's the same as --num-trials. """, ) args.add_argument( "--backend", type=str, choices=["graph", "vm"], default="graph", help="The backend to use for relay compilation(graph / vm).", ) # TODO(@yelite): Add a layout arg to transform the network after # ingesting into Relay and before feeding into MetaSchedule. # Evaluator-related config args.add_argument( "--number", type=int, default=3, help="The number of times to run the model for taking average in a single measurement.", ) args.add_argument( "--repeat", type=int, default=1, help="The number of times to repeat the measurement.", ) args.add_argument( "--min-repeat-ms", type=int, default=100, help=""" Minimum repeat time in ms. The number of runs will be increased if the actual repeat time is lowered than this. """, ) args.add_argument( "--adaptive-training", action="store_true", help="Whether to use adaptive training for cost model.", ) args.add_argument( "--cpu-flush", action="store_true", help="Whether to perform CPU cache flush.", ) # RPC-related args args.add_argument( "--rpc-host", type=str, help="Host of the RPC Tracker for tuning. Use LocalRunner if not provided", ) args.add_argument( "--rpc-port", type=int, help="Port of the RPC Tracker for tuning", ) args.add_argument( "--rpc-key", type=str, help="Key of the RPC Tracker for tuning", ) parsed = args.parse_args() if parsed.disallowed_op == "all": disallowed_op = set(DisallowedOperator) else: disallowed_op = {DisallowedOperator(v) for v in parsed.disallowed_op.split(",")} parsed.disallowed_op = disallowed_op # Trim all args, otherwise it confuses the arg parser of timm_efficientdet sys.argv = sys.argv[:1] return parsed logging.basicConfig( format="%(asctime)s.%(msecs)03d %(levelname)s %(message)s", datefmt="%Y-%m-%d %H:%M:%S", ) logging.getLogger("tvm.meta_schedule").setLevel(logging.DEBUG) ARGS = parse_args() IS_CUDA = ARGS.target.kind.name == "cuda" logger = logging.getLogger(__name__) # pylint: disable=invalid-name logger.setLevel(logging.INFO) runner = load_torchdynamo_benchmark_runner( # pylint: disable=invalid-name IS_CUDA, cosine_similarity=ARGS.result_metric == ResultComparisonMetric.COSINE, float32=ARGS.float32, disallowed_operators=ARGS.disallowed_op, ) def get_meta_schedule_runner() -> ms.runner.PyRunner: """ Get the Runner for MetaSchedule. It returns RPCRunner if --rpc-host is given, otherwise it returns LocalRunner """ if ARGS.rpc_host is not None: assert ARGS.rpc_port is not None, "Missing rpc_port" assert ARGS.rpc_key is not None, "Missing rpc_key" return ms.runner.RPCRunner( rpc_config=ms.runner.RPCConfig( tracker_host=ARGS.rpc_host, tracker_port=ARGS.rpc_port, tracker_key=ARGS.rpc_key, session_timeout_sec=600, ), evaluator_config=ms.runner.EvaluatorConfig( number=ARGS.number, repeat=ARGS.repeat, min_repeat_ms=ARGS.min_repeat_ms, enable_cpu_cache_flush=ARGS.cpu_flush, ), alloc_repeat=1, ) else: warnings.warn("Falling back to MetaSchedule LocalRunner because --rpc-host isn't provided.") return ms.runner.LocalRunner() def get_graph_executor_forward( graph_executor_factory: tvm.runtime.Module, device: tvm.runtime.Device ) -> Callable: """ Get the forward function for graph executor, in order to integrate with TorchDynamo. """ # It has to lazily import this package, loading the C++ PyTorch integration # after the transformers package is imported when loading model. Otherwise # there will be segfault caused by the protobuf library. import tvm.contrib.torch # pylint: disable=import-outside-toplevel, unused-import, redefined-outer-name save_runtime_mod = get_global_func("tvmtorch.save_runtime_mod", allow_missing=True) if save_runtime_mod is None: warnings.warn( "C++ PyTorch TVM integration is missing. Fallback to Python forward function." "Build TVM with 'USE_PT_TVMDSOOP' to enable the C++ custom operator" ) mod = GraphModule(graph_executor_factory["default"](device)) def forward(*args): if IS_CUDA: torch.cuda.synchronize() args = tuple(arg.detach().contiguous() for arg in args) for idx, arg in enumerate(args, 0): mod.set_input( f"inp_{idx}", tvm.nd.from_dlpack(arg), ) mod.run() device.sync() result = [torch.from_dlpack(mod.get_output(i)) for i in range(mod.get_num_outputs())] return result return forward else: save_runtime_mod(graph_executor_factory.module) module = torch.classes.tvm_torch.GraphExecutorFactoryWrapper() def forward(*args): # type: ignore # isort: skip, pylint: disable=function-redefined return module.forward(args) return forward def get_vm_forward(virtual_machine: VirtualMachine, device: tvm.runtime.Device) -> Callable: """ Get the forward function for VM, in order to integrate with TorchDynamo. """ def forward(*args): if IS_CUDA: torch.cuda.synchronize() args = tuple(tvm.nd.from_dlpack(arg.detach().contiguous()) for arg in args) result = virtual_machine.invoke("main", *args) device.sync() if isinstance(result, tvm.nd.NDArray): result = [result] return [torch.from_dlpack(m) for m in result] return forward def should_skip_subgraph(graph_module: torch.fx.GraphModule) -> bool: """ Returns whether it should skip optimizing the input graph module. The graph could be empyt or only containing nodes calling function for side effect. """ graph = graph_module.graph inputs = [n for n in graph.nodes if n.op == "placeholder"] outputs = [n for n in graph.nodes if n.op == "output"] return len(inputs) == 0 and all(output.args == ((),) for output in outputs) def create_tvm_task_collection_backend() -> Tuple[Callable, List[ms.ExtractedTask]]: """ This torchdynamo backend only collects the extracted tasks from MetaSchedule. It doesn't tune the model. """ subgraph_idx = 0 subgraphs_dir = os.path.join(ARGS.work_dir, "subgraphs") os.makedirs(subgraphs_dir, exist_ok=True) collected_tasks = [] task_index: Dict[int, List[ms.ExtractedTask]] = defaultdict(list) def collect_task(task): task_hash = tvm.ir.structural_hash(task.dispatched[0]) for duplicate_task in task_index[task_hash]: if tvm.ir.structural_equal(duplicate_task.dispatched[0], task.dispatched[0]): duplicate_task.weight += task.weight return task_index[task_hash].append(task) collected_tasks.append(task) def backend(graph_module, example_inputs): nonlocal subgraph_idx torch.save(graph_module, os.path.join(subgraphs_dir, f"graph_module_{subgraph_idx}")) torch.save(example_inputs, os.path.join(subgraphs_dir, f"example_inputs_{subgraph_idx}")) if should_skip_subgraph(graph_module): return graph_module.forward jit_mod = torch.jit.trace(graph_module, example_inputs) shape_list = [(f"inp_{idx}", i.shape) for idx, i in enumerate(example_inputs)] ir_mod, params = tvm.relay.frontend.from_pytorch(jit_mod, shape_list) extracted_tasks = ms.relay_integration.extract_tasks( mod=ir_mod, target=ARGS.target, params=params, ) old_tasks_count = len(collected_tasks) for task in extracted_tasks: collect_task(task) logger.info( "Extracted %d tasks from graph %d, with %d new tasks", len(extracted_tasks), subgraph_idx, len(collected_tasks) - old_tasks_count, ) subgraph_idx += 1 return graph_module.forward return backend, collected_tasks def create_tvm_compilation_backend(database: ms.database.Database) -> Callable: """ This torchdynamo backend compiles the model using history best record from the MetaSchedule database. """ def backend(graph_module, example_inputs): if should_skip_subgraph(graph_module): return graph_module.forward jit_mod = torch.jit.trace(graph_module, example_inputs) shape_list = [(f"inp_{idx}", i.shape) for idx, i in enumerate(example_inputs)] ir_mod, params = tvm.relay.frontend.from_pytorch(jit_mod, shape_list) lib = ms.relay_integration.compile_relay( database=database, mod=ir_mod, target=ARGS.target, params=params, backend=ARGS.backend, ) device = tvm.cuda(0) if IS_CUDA else tvm.cpu(0) if ARGS.backend == "graph": return get_graph_executor_forward(lib, device) elif ARGS.backend == "vm": vm = VirtualMachine(lib, device) # pylint: disable=invalid-name return get_vm_forward(vm, device) else: raise RuntimeError(f"Unknown backend {ARGS.backend}") return backend def format_time(seconds: float) -> str: """ Format elapsed time based on its value. """ if seconds > 1: return f"{seconds:.3g}s" else: return f"{seconds * 1000:.3g}ms" def is_output_correct(output: torch.Tensor, expected: torch.Tensor) -> bool: """ Check whether the output is correct. """ comparison_metric = ARGS.result_metric if comparison_metric == ResultComparisonMetric.COSINE: return same(expected, output, cosine_similarity=True) elif comparison_metric == ResultComparisonMetric.ALLCLOSE: return same(expected, output, tol=1e-4) else: raise RuntimeError(f"Unknown comparison metric {comparison_metric}") def inspect_output_error(output, expected): """ Inpsect the error between the actual output and expected output. """ if not isinstance(output, torch.Tensor): logger.info( f"Unsupported type for error inspection: {type(output).__name__}." f"Please manually check output.pt" ) return output = output.cpu().float() expected = expected.cpu().float() abs_error = (output - expected).abs() rel_error = (abs_error / expected).abs() def format_error_table(error, bins) -> str: bin_tensor = torch.as_tensor([float(b) for b in bins], dtype=error.dtype) error_hist = torch.histogram(error, bin_tensor).hist.int() return "\n".join(f"< {b}\t{e}" for e, b in zip(error_hist, bins[1:])) abs_error_bins = [ "-1e10", "0", "1e-8", "1e-6", "1e-5", "1e-4", "1e-3", "1e-2", "1e-1", "1", "1e10", ] rel_error_bins = [ "-1e10", "0", "1e-4", "1e-3", "1e-2", "1e-1", "1", "1e1", "1e2", "1e3", "1e100", ] large_rel_error_idx = rel_error > 1 abs_error_with_large_rel_error = abs_error[large_rel_error_idx] logger.error(f"Expected (PyTorch eager): {expected}") logger.error(f"Actual (Optimized): {output}") logger.error(f"Absolute Error\n{format_error_table(abs_error, abs_error_bins)}") logger.error(f"Relative Error\n{format_error_table(rel_error, rel_error_bins)}") logger.error( f"Max absolute error for position with large relative error (> 1):" f"{abs_error_with_large_rel_error.max()}" ) def performance_experiment( model_iter_fn: Callable, model: torch.nn.Module, example_inputs: Tuple[torch.Tensor], ) -> str: """ Performs the actual benchmarking Simplified from https://github.com/pytorch/torchdynamo/blob/c537639f9712621dc04ca09908796dbbe86c354b/benchmarks/common.py#L494 pylint: disable=line-too-long """ timings = np.zeros((ARGS.benchmark_repeat, 2), np.float64) if IS_CUDA: torch.cuda.empty_cache() is_correct = True frozen_model_iter_fn = torchdynamo.run(model_iter_fn) for _ in range(ARGS.benchmark_warmup_rounds): frozen_model_iter_fn(model, example_inputs) model_iter_fn(model, example_inputs) for rep in range(ARGS.benchmark_repeat): # interleave the runs to handle frequency scaling and load changes timings[rep, 0], expected_output = timed( model, model_iter_fn, example_inputs, return_result=True ) timings[rep, 1], actual_output = timed( model, frozen_model_iter_fn, example_inputs, return_result=True ) is_correct = is_correct and is_output_correct(expected_output, actual_output) pvalue = ttest_ind(timings[:, 0], timings[:, 1]).pvalue median = np.median(timings, axis=0) speedup = median[0] / median[1] logger.info( f"eager:{format_time(median[0])} " f"optimized:{format_time(median[1])} " f"speedup:{speedup:.3f}x p:{pvalue:.3f}" ) torch.save(actual_output, os.path.join(ARGS.work_dir, "output.pt")) torch.save(expected_output, os.path.join(ARGS.work_dir, "expected.pt")) if not is_correct: logger.error("Result is incorrect.") inspect_output_error(actual_output, expected_output) return "" def get_torch_device_type(target: tvm.target.Target) -> str: if target.kind.name == "llvm": return "cpu" elif target.kind.name == "cuda": return "cuda" else: raise RuntimeError(f"Unsupported target {target}") def main(): """ Entry point of the benchmark """ describe() meta_schedule_work_dir = os.path.join(ARGS.work_dir, "meta_schedule") os.makedirs(meta_schedule_work_dir, exist_ok=True) database = ms.database.JSONDatabase(work_dir=meta_schedule_work_dir) if not ARGS.mode.should_tune: if len(database) == 0: raise RuntimeError( "Script is running in eval mode while the tuning database is empty. " "Please tune the model first." ) if IS_CUDA and ARGS.cpu_flush: warnings.warn( "Benchmark is running on CUDA, while --cpu-flush is turned on. " "This flag will have no effect on CUDA." ) ARGS.cpu_flush = False try: logger.info(f"Loading model with batch size: {ARGS.batch_size}") _, name, model, example_inputs, batch_size = runner.load_model( get_torch_device_type(ARGS.target), ARGS.model, batch_size=ARGS.batch_size, ) model, example_inputs = runner.maybe_cast(model, example_inputs) logger.info(f"Got model with batch size: {batch_size}") except NotImplementedError: logger.exception(f"{ARGS.model} failed to load") raise with contextlib.ExitStack() as stack: profiler = stack.enter_context(ms.Profiler()) stack.enter_context(torch.no_grad()) tasks_path = os.path.join(ARGS.work_dir, "extracted_tasks") if ARGS.mode.should_extract: task_collect_backend, extracted_tasks = create_tvm_task_collection_backend() task_collect_ctx = torchdynamo.optimize(task_collect_backend) task_collect_ctx(runner.model_iter_fn)(model, example_inputs) with open(tasks_path, "wb") as f: pickle.dump(extracted_tasks, f) else: with open(tasks_path, "rb") as f: extracted_tasks = pickle.load(f) if ARGS.mode.should_tune: tasks, task_weights = ms.relay_integration.extracted_tasks_to_tune_contexts( extracted_tasks=extracted_tasks, work_dir=ARGS.work_dir, strategy=ARGS.strategy, ) database = ms.tune.tune_tasks( tasks=tasks, task_weights=task_weights, work_dir=ARGS.work_dir, max_trials_global=ARGS.num_trials, max_trials_per_task=ARGS.max_trials_per_task, runner=get_meta_schedule_runner(), # type: ignore database=database, cost_model=ms.cost_model.XGBModel( # type: ignore extractor=ms.feature_extractor.PerStoreFeature(), adaptive_training=ARGS.adaptive_training, ), ) if ARGS.mode.should_eval: torchdynamo.reset() model_compile_ctx = torchdynamo.optimize(create_tvm_compilation_backend(database)) model_compile_ctx(runner.model_iter_fn)(model, example_inputs) with torch.no_grad(): performance_experiment(runner.model_iter_fn, model, example_inputs) print(profiler.table()) if __name__ == "__main__": main()
25,562
31.276515
160
py
tvm
tvm-main/python/tvm/meta_schedule/testing/torchbench/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License.
785
45.235294
62
py
tvm
tvm-main/python/tvm/meta_schedule/measure_callback/add_to_database.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """A callback that adds the measurement results into the database""" from tvm._ffi import register_object from .. import _ffi_api from .measure_callback import MeasureCallback @register_object("meta_schedule.AddToDatabase") class AddToDatabase(MeasureCallback): def __init__(self) -> None: """A callback that adds the measurement results into the database""" self.__init_handle_by_constructor__( _ffi_api.MeasureCallbackAddToDatabase, # type: ignore # pylint: disable=no-member )
1,309
41.258065
94
py
tvm
tvm-main/python/tvm/meta_schedule/measure_callback/remove_build_artifact.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """A callback that removes the build artifacts from the disk""" from tvm._ffi import register_object from .. import _ffi_api from .measure_callback import MeasureCallback @register_object("meta_schedule.RemoveBuildArtifact") class RemoveBuildArtifact(MeasureCallback): def __init__(self) -> None: """A callback that removes the build artifacts from the disk""" self.__init_handle_by_constructor__( _ffi_api.MeasureCallbackRemoveBuildArtifact, # type: ignore # pylint: disable=no-member )
1,317
41.516129
100
py
tvm
tvm-main/python/tvm/meta_schedule/measure_callback/measure_callback.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Meta Schedule MeasureCallback.""" from typing import TYPE_CHECKING, Callable, List, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from .. import _ffi_api from ..builder import BuilderResult from ..runner import RunnerResult from ..search_strategy import MeasureCandidate from ..utils import _get_default_str if TYPE_CHECKING: from ..task_scheduler import TaskScheduler @register_object("meta_schedule.MeasureCallback") class MeasureCallback(Object): """Rules to apply after measure results is available.""" CallbackListType = Union[List["MeasureCallback"], "MeasureCallback", Literal["default"]] def apply( self, task_scheduler: "TaskScheduler", task_id: int, measure_candidates: List[MeasureCandidate], builder_results: List[BuilderResult], runner_results: List[RunnerResult], ) -> None: """Apply a measure callback to the given schedule. Parameters ---------- task_scheduler: TaskScheduler The task scheduler. task_id: int The task id. measure_candidates: List[MeasureCandidate] The measure candidates. builder_results: List[BuilderResult] The builder results by building the measure candidates. runner_results: List[RunnerResult] The runner results by running the built measure candidates. """ return _ffi_api.MeasureCallbackApply( # type: ignore # pylint: disable=no-member self, task_scheduler, task_id, measure_candidates, builder_results, runner_results, ) @staticmethod def create(kind: Literal["default"]) -> List["MeasureCallback"]: """Create a list of measure callbacks.""" if kind == "default": return _ffi_api.MeasureCallbackDefault() # type: ignore # pylint: disable=no-member raise ValueError(f"Unknown kind of MeasureCallback list: {kind}") @register_object("meta_schedule.PyMeasureCallback") class _PyMeasureCallback(MeasureCallback): """ A TVM object measure callback to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyMeasureCallback """ def __init__(self, f_apply: Callable, f_as_string: Callable = None): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.MeasureCallbackPyMeasureCallback, # type: ignore # pylint: disable=no-member f_apply, f_as_string, ) class PyMeasureCallback: """ An abstract measure callback with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyMeasureCallback, "methods": ["apply", "__str__"], } def apply( self, task_scheduler: "TaskScheduler", task_id: int, measure_candidates: List[MeasureCandidate], builder_results: List[BuilderResult], runner_results: List[RunnerResult], ) -> None: """Apply a measure callback to the given schedule. Parameters ---------- task_scheduler: TaskScheduler The task scheduler. task_id: int The task id. measure_candidates: List[MeasureCandidate] The measure candidates. builder_results: List[BuilderResult] The builder results by building the measure candidates. runner_results: List[RunnerResult] The runner results by running the built measure candidates. """ raise NotImplementedError def __str__(self) -> str: return _get_default_str(self)
4,756
32.034722
98
py
tvm
tvm-main/python/tvm/meta_schedule/measure_callback/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """The tvm.meta_schedule.measure_callback package.""" from .add_to_database import AddToDatabase from .measure_callback import MeasureCallback, PyMeasureCallback from .remove_build_artifact import RemoveBuildArtifact from .update_cost_model import UpdateCostModel
1,049
46.727273
64
py
tvm
tvm-main/python/tvm/meta_schedule/measure_callback/update_cost_model.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """A measure callback that updates the cost model""" from tvm._ffi import register_object from .. import _ffi_api from .measure_callback import MeasureCallback @register_object("meta_schedule.UpdateCostModel") class UpdateCostModel(MeasureCallback): def __init__(self) -> None: """A measure callback that updates the cost model""" self.__init_handle_by_constructor__( _ffi_api.MeasureCallbackUpdateCostModel, # type: ignore # pylint: disable=no-member )
1,283
40.419355
96
py
tvm
tvm-main/python/tvm/meta_schedule/feature_extractor/feature_extractor.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Meta Schedule FeatureExtractor.""" from typing import Callable, List, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from tvm.runtime.ndarray import NDArray from .. import _ffi_api from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext from ..utils import _get_default_str @register_object("meta_schedule.FeatureExtractor") class FeatureExtractor(Object): """Extractor for features from measure candidates for use in cost model.""" FeatureExtractorType = Union[Literal["per-store-feature"], "FeatureExtractor"] def extract_from( self, context: TuneContext, candidates: List[MeasureCandidate] ) -> List[NDArray]: """Extract features from the given measure candidate. Parameters ---------- context : TuneContext The tuning context for feature extraction. candidates : List[MeasureCandidate] The measure candidates to extract features from. Returns ------- features : List[NDArray] The feature tvm ndarray extracted. """ result = _ffi_api.FeatureExtractorExtractFrom( # type: ignore # pylint: disable=no-member self, context, candidates ) return result @staticmethod def create( kind: Literal["per-store-feature"], *args, **kwargs, ) -> "FeatureExtractor": """Create a CostModel.""" from . import PerStoreFeature # pylint: disable=import-outside-toplevel if kind == "per-store-feature": return PerStoreFeature(*args, **kwargs) # type: ignore raise ValueError(f"Unknown CostModel: {kind}") @register_object("meta_schedule.PyFeatureExtractor") class _PyFeatureExtractor(FeatureExtractor): """ A TVM object feature extractor to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyFeatureExtractor """ def __init__(self, f_extract_from: Callable, f_as_string: Callable = None): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.FeatureExtractorPyFeatureExtractor, # type: ignore # pylint: disable=no-member f_extract_from, f_as_string, ) class PyFeatureExtractor: """ An abstract feature extractor with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyFeatureExtractor, "methods": ["extract_from", "__str__"], } def extract_from( self, context: TuneContext, candidates: List[MeasureCandidate] ) -> List[NDArray]: """Extract features from the given measure candidate. Parameters ---------- context : TuneContext The tuning context for feature extraction. candidates : List[MeasureCandidate] The measure candidates to extract features from. Returns ------- features : List[NDArray] The feature tvm ndarray extracted. """ raise NotImplementedError def __str__(self) -> str: return _get_default_str(self)
4,206
31.361538
100
py
tvm
tvm-main/python/tvm/meta_schedule/feature_extractor/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.feature_extractor package. Meta Schedule feature extractors that extracts features from measure candidates for use in cost model. """ from .feature_extractor import FeatureExtractor, PyFeatureExtractor from .per_store_feature import PerStoreFeature from .random_feature_extractor import RandomFeatureExtractor
1,121
43.88
67
py
tvm
tvm-main/python/tvm/meta_schedule/feature_extractor/random_feature_extractor.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Random Feature Extractor.""" from typing import List, Tuple, Union import numpy as np # type: ignore from tvm.runtime.ndarray import NDArray, array from ..feature_extractor import PyFeatureExtractor from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext from ..utils import derived_object @derived_object class RandomFeatureExtractor(PyFeatureExtractor): """Random Feature Extractor Parameters ---------- feature_size : int The size of each block's feature vector. max_block_num : int The maximum number of blocks in each schedule. random_state : Union[Tuple[str, np.ndarray, int, int, float], dict] The current random state of the f """ feature_size: int max_block_num: int random_state: Union[Tuple[str, np.ndarray, int, int, float], dict] def __init__(self, *, feature_size: int = 30, max_block_num: int = 5, seed=0): super().__init__() assert max_block_num >= 1, "Max block number must be greater or equal to one!" self.max_block_num = max_block_num self.feature_size = feature_size np.random.seed(seed) self.random_state = np.random.get_state() def extract_from( self, context: TuneContext, candidates: List[MeasureCandidate] ) -> List[NDArray]: np.random.set_state(self.random_state) result = [ np.random.rand(np.random.randint(1, self.max_block_num + 1), self.feature_size) for candidate in candidates ] self.random_state = np.random.get_state() return [array(x) for x in result]
2,411
36.107692
91
py
tvm
tvm-main/python/tvm/meta_schedule/feature_extractor/per_store_feature.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name """We extract one feature vector per BufferStoreNode statement in a TIR Stmt, so we call this feature as "per-store" feature. """ from tvm._ffi import register_object from .. import _ffi_api from .feature_extractor import FeatureExtractor @register_object("meta_schedule.PerStoreFeature") class PerStoreFeature(FeatureExtractor): """PerStoreFeature extracts one feature vector per BufferStoreNode Parameters ---------- buffers_per_store : int The number of buffers in each BufferStore; Pad or truncate if necessary. arith_intensity_curve_num_samples : int The number of samples used in the arithmetic intensity curve. cache_line_bytes : int The number of bytes in a cache line. extract_workload : bool Whether to extract features in the workload in tuning context or not. """ buffers_per_store: int """The number of buffers in each BufferStore; Pad or truncate if necessary.""" arith_intensity_curve_num_samples: int # pylint: disable=invalid-name """The number of samples used in the arithmetic intensity curve.""" cache_line_bytes: int """The number of bytes in a cache line.""" extract_workload: bool """Whether to extract features in the workload in tuning context or not.""" feature_vector_length: int """Length of the feature vector.""" def __init__( self, buffers_per_store: int = 5, arith_intensity_curve_num_samples: int = 10, cache_line_bytes: int = 64, extract_workload: bool = False, ): self.__init_handle_by_constructor__( _ffi_api.FeatureExtractorPerStoreFeature, # type: ignore # pylint: disable=no-member buffers_per_store, arith_intensity_curve_num_samples, cache_line_bytes, extract_workload, )
2,665
38.205882
97
py
tvm
tvm-main/python/tvm/meta_schedule/task_scheduler/task_scheduler.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Auto-tuning Task Scheduler""" from typing import Callable, List, Optional, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from .. import _ffi_api from ..builder import Builder, BuilderResult from ..cost_model import CostModel from ..database import Database from ..logging import get_logger, get_logging_func from ..measure_callback import MeasureCallback from ..runner import Runner, RunnerResult from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext logger = get_logger(__name__) # pylint: disable=invalid-name @register_object("meta_schedule.TaskRecord") class TaskRecord(Object): """The running record of a task.""" ctx: TuneContext task_weight: float flop: float is_terminated: bool build_error_count: int run_error_count: int measure_candidates: List[MeasureCandidate] builder_results: List[BuilderResult] runner_results: List[RunnerResult] @register_object("meta_schedule.TaskScheduler") class TaskScheduler(Object): """The abstract task scheduler interface.""" tasks_: List[TaskRecord] measure_callbacks_: List[MeasureCallback] database_: Optional[Database] cost_model_: Optional[CostModel] remaining_tasks_: int TaskSchedulerType = Union["TaskScheduler", Literal["gradient", "round-robin"]] def next_task_id(self) -> int: """Fetch the next task id. Returns ------- next_task_id : int The next task id. """ return _ffi_api.TaskSchedulerNextTaskId(self) # type: ignore # pylint: disable=no-member def join_running_task(self, task_id: int) -> List[RunnerResult]: """Wait until the task is finished. Parameters ---------- task_id : int The task id to be joined. Returns ------- results : List[RunnerResult] The list of results. """ return _ffi_api.TaskSchedulerJoinRunningTask(self, task_id) # type: ignore # pylint: disable=no-member def tune( self, tasks: List[TuneContext], task_weights: List[float], max_trials_global: int, max_trials_per_task: int, num_trials_per_iter: int, builder: Builder, runner: Runner, measure_callbacks: List[MeasureCallback], database: Optional[Database], cost_model: Optional[CostModel], ) -> None: """Auto-tuning. Parameters ---------- tasks : List[TuneContext] The list of tuning contexts as tasks. task_weights : List[float] The list of task weights. max_trials_global : int The maximum number of trials globally. max_trials_per_task : int The maximum number of trials per task. num_trials_per_iter : int The number of trials per iteration. builder : Builder The builder. runner : Runner The runner. measure_callbacks : List[MeasureCallback] The list of measure callbacks. database : Optional[Database] The database. cost_model : Optional[CostModel] The cost model. """ task_weights = [float(w) for w in task_weights] _ffi_api.TaskSchedulerTune( # type: ignore # pylint: disable=no-member self, tasks, task_weights, max_trials_global, max_trials_per_task, num_trials_per_iter, builder, runner, measure_callbacks, database, cost_model, ) def terminate_task(self, task_id: int) -> None: """Terminate the task Parameters ---------- task_id : int The task id to be terminated. """ _ffi_api.TaskSchedulerTerminateTask(self, task_id) # type: ignore # pylint: disable=no-member def touch_task(self, task_id: int) -> None: """Touch the task and update its status Parameters ---------- task_id : int The task id to be checked. """ _ffi_api.TaskSchedulerTouchTask(self, task_id) # type: ignore # pylint: disable=no-member def print_tuning_statistics(self) -> None: """Print out a human-readable format of the tuning statistics.""" return _ffi_api.TaskSchedulerPrintTuningStatistics(self) # type: ignore # pylint: disable=no-member @staticmethod def create( # pylint: disable=keyword-arg-before-vararg kind: Literal["round-robin", "gradient"] = "gradient", *args, **kwargs, ) -> "TaskScheduler": """Create a task scheduler.""" from . import ( # pylint: disable=import-outside-toplevel GradientBased, RoundRobin, ) if kind == "round-robin": return RoundRobin(*args, **kwargs) # type: ignore if kind == "gradient": return GradientBased(*args, **kwargs) raise ValueError(f"Unknown TaskScheduler name: {kind}") create = TaskScheduler.create # pylint: disable=invalid-name @register_object("meta_schedule.PyTaskScheduler") class _PyTaskScheduler(TaskScheduler): """ A TVM object task scheduler to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyTaskScheduler """ def __init__( self, f_next_task_id: Callable, f_join_running_task: Callable, f_tune: Callable, ): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.TaskSchedulerPyTaskScheduler, # type: ignore # pylint: disable=no-member get_logging_func(logger), f_next_task_id, f_join_running_task, f_tune, ) class PyTaskScheduler: """ An abstract task scheduler with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyTaskScheduler, "fields": [], "methods": ["next_task_id", "join_running_task", "tune"], } def __init__(self): ... def tune( self, tasks: List[TuneContext], task_weights: List[float], max_trials_global: int, max_trials_per_task: int, builder: Builder, runner: Runner, measure_callbacks: List[MeasureCallback], database: Optional[Database], cost_model: Optional[CostModel], ) -> None: """Auto-tuning.""" # Using self._outer to replace the self pointer _ffi_api.TaskSchedulerTune( # type: ignore # pylint: disable=no-member self._outer(), # type: ignore # pylint: disable=no-member tasks, task_weights, max_trials_global, max_trials_per_task, builder, runner, measure_callbacks, database, cost_model, ) def next_task_id(self) -> int: """Fetch the next task id. Returns ------- next_task_id : int The next task id. """ raise NotImplementedError def join_running_task(self, task_id: int) -> List[RunnerResult]: """Wait until the task is finished. Parameters ---------- task_id : int The task id to be joined. """ # Using self._outer to replace the self pointer return _ffi_api.TaskSchedulerJoinRunningTask(self._outer(), task_id) # type: ignore # pylint: disable=no-member
8,661
29.716312
120
py
tvm
tvm-main/python/tvm/meta_schedule/task_scheduler/round_robin.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Round Robin Task Scheduler""" from tvm._ffi import register_object from .. import _ffi_api from ..logging import get_logger, get_logging_func from .task_scheduler import TaskScheduler logger = get_logger(__name__) # pylint: disable=invalid-name @register_object("meta_schedule.RoundRobin") class RoundRobin(TaskScheduler): """Round Robin Task Scheduler""" def __init__(self) -> None: """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.TaskSchedulerRoundRobin, # type: ignore # pylint: disable=no-member get_logging_func(logger), )
1,396
36.756757
89
py
tvm
tvm-main/python/tvm/meta_schedule/task_scheduler/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.task_scheduler package. Meta Schedule task scheduler that manage the task scheduling for measure candidates generation and measurement, then save records to the database. """ from .gradient_based import GradientBased from .round_robin import RoundRobin from .task_scheduler import PyTaskScheduler, TaskScheduler, create
1,131
42.538462
66
py
tvm
tvm-main/python/tvm/meta_schedule/task_scheduler/gradient_based.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Gradient Based Task Scheduler""" from tvm._ffi import register_object from .. import _ffi_api from ..logging import get_logger, get_logging_func from .task_scheduler import TaskScheduler logger = get_logger(__name__) # pylint: disable=invalid-name @register_object("meta_schedule.GradientBased") class GradientBased(TaskScheduler): """Gradient Based Task Scheduler""" def __init__( self, *, alpha: float = 0.2, window_size: int = 3, seed: int = -1, ) -> None: """Constructor. Parameters ---------- alpha : float = 0.2 The parameter alpha in gradient computation. window_size : int = 3 The parameter to control backward window size in gradient computation. seed : int = -1 The random seed. """ self.__init_handle_by_constructor__( _ffi_api.TaskSchedulerGradientBased, # type: ignore # pylint: disable=no-member get_logging_func(logger), alpha, window_size, seed, )
1,880
32.589286
92
py
tvm
tvm-main/python/tvm/meta_schedule/builder/local_builder.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Local builder that compile on the local host""" import os import tempfile from typing import Callable, Dict, List, Optional, Union from tvm._ffi import register_func from tvm.ir import IRModule from tvm.runtime import Module, NDArray, load_param_dict, save_param_dict from tvm.target import Target from ...contrib.popen_pool import MapResult, PopenPoolExecutor, StatusKind from ..logging import get_logger from ..utils import cpu_count, derived_object, get_global_func_with_default_on_worker from .builder import BuilderInput, BuilderResult, PyBuilder logger = get_logger(__name__) # pylint: disable=invalid-name T_BUILD = Callable[ # pylint: disable=invalid-name [IRModule, Target, Optional[Dict[str, NDArray]]], Module ] T_EXPORT = Callable[[Module], str] # pylint: disable=invalid-name def _serialize_params(params: Optional[Dict[str, NDArray]]) -> Optional[bytearray]: if params is None: return None return save_param_dict(params) def _deserialize_params(params: Optional[bytearray]) -> Optional[Dict[str, NDArray]]: if params is None: return None return load_param_dict(params) @derived_object class LocalBuilder(PyBuilder): """A builder that builds the given input on local host. Parameters ---------- pool : PopenPoolExecutor The process pool to run the build. max_workers: int The max number of Popen workers. timeout_sec : float The timeout in seconds for the build. initializer: Optional[Callable[[], None]] The initializer function for each popen worker. f_build : Union[None, str, T_BUILD] Name of the build function to be used. Defaults to `meta_schedule.builder.default_build`. f_export : Union[None, str, T_EXPORT] Name of the export function to be used. Defaults to `meta_schedule.builder.default_export`. Attributes ---------- T_BUILD : typing._GenericAlias The signature of the function `f_build`, which is .. code-block:: python def default_build( mod: IRModule, target: Target, params: Optional[Dict[str, NDArray]] ) -> Module: ... T_EXPORT : typing._GenericAlias The signature of the function `f_export`, which is .. code-block:: python def default_export(mod: Module) -> str: ... Note ---- The build function and export function should be registered in the worker process. The worker process is only aware of functions registered in TVM package, if there are extra functions to be registered, please send the registration logic via initializer. """ max_workers: int timeout_sec: float initializer: Optional[Callable[[], None]] f_build: Union[None, str, T_BUILD] f_export: Union[None, str, T_EXPORT] def __init__( self, *, max_workers: Optional[int] = None, timeout_sec: float = 30.0, f_build: Union[None, str, T_BUILD] = None, f_export: Union[None, str, T_EXPORT] = None, initializer: Optional[Callable[[], None]] = None, ) -> None: """Constructor. Parameters ---------- max_workers : Optional[int] The maximum number of worker processes to be used. Defaults to number of CPUs. timeout_sec : float The timeout in seconds for the build. f_build : T_BUILD Name of the build function to be used. Defaults to `meta_schedule.builder.default_build`. f_export : T_EXPORT Name of the export function to be used. Defaults to `meta_schedule.builder.default_export`. initializer : Optional[Callable[[], None]] The initializer to be used for the worker processes. """ super().__init__() if max_workers is None: max_workers = cpu_count(logical=True) logger.info("LocalBuilder: max_workers = %d", max_workers) self.max_workers = max_workers self.timeout_sec = timeout_sec self.initializer = initializer self.f_build = f_build self.f_export = f_export self._sanity_check() def build(self, build_inputs: List[BuilderInput]) -> List[BuilderResult]: results: List[BuilderResult] = [] map_result: MapResult # Here we restart the PopenPool everytime because of a known memory leak issue with the # PopenPool workers after a couple times of usage. We don't apply the same to runners to # avoid potential problem caused by async behaviour. pool = PopenPoolExecutor( max_workers=self.max_workers, timeout=self.timeout_sec, initializer=self.initializer, ) # Dispatch the build inputs to the worker processes. for map_result in pool.map_with_error_catching( lambda x: _worker_func(*x), [ ( self.f_build, self.f_export, build_input.mod, build_input.target, _serialize_params(build_input.params), ) for build_input in build_inputs ], ): if map_result.status == StatusKind.COMPLETE: results.append(BuilderResult(map_result.value, None)) elif map_result.status == StatusKind.TIMEOUT: results.append( BuilderResult( None, f"LocalBuilder: Timeout, killed after {self.timeout_sec} seconds", ) ) elif map_result.status == StatusKind.EXCEPTION: results.append( BuilderResult( None, "LocalBuilder: An exception occurred\n" + str(map_result.value), ) ) else: raise ValueError("Unreachable: unexpected result: {map_result}") del pool return results def _sanity_check(self) -> None: def _check(f_build, f_export) -> None: get_global_func_with_default_on_worker(name=f_build, default=None) get_global_func_with_default_on_worker(name=f_export, default=None) # Same reason for the single use PopenPool as mentioned above pool = PopenPoolExecutor( max_workers=self.max_workers, timeout=self.timeout_sec, initializer=self.initializer, ) value = pool.submit(_check, self.f_build, self.f_export) value.result() del pool def _worker_func( _f_build: Union[None, str, T_BUILD], _f_export: Union[None, str, T_EXPORT], mod: IRModule, target: Target, params: Optional[bytearray], ) -> str: # Step 0. Get the registered functions f_build: T_BUILD = get_global_func_with_default_on_worker( _f_build, default_build, ) f_export: T_EXPORT = get_global_func_with_default_on_worker( _f_export, default_export, ) # Step 1. Build the IRModule rt_mod: Module = f_build(mod, target, _deserialize_params(params)) # Step 2. Export the Module artifact_path: str = f_export(rt_mod) return artifact_path @register_func("meta_schedule.builder.default_build") def default_build(mod: IRModule, target: Target, _params: Optional[Dict[str, NDArray]]) -> Module: """Default build function. Parameters ---------- mod : IRModule The IRModule to be built. target : Target The target to be built. _params : Optional[Dict[str, NDArray]] The parameters to be used for the build. Must be None. Returns ------- rt_mod : Module The built Module. """ # pylint: disable=import-outside-toplevel from tvm.driver import build as tvm_build from tvm.tir.transform import RemoveWeightLayoutRewriteBlock # pylint: enable=import-outside-toplevel mod = RemoveWeightLayoutRewriteBlock(skip_ndarray_rewrite=True)(mod) return tvm_build(mod, target=target) @register_func("meta_schedule.builder.default_export") def default_export(mod: Module) -> str: """Default export function. Parameters ---------- mod : Module The Module to be exported. Returns ------- artifact_path : str The path to the exported Module. """ from tvm.contrib.tar import tar # pylint: disable=import-outside-toplevel artifact_path = os.path.join(tempfile.mkdtemp(), "tvm_tmp_mod." + tar.output_format) mod.export_library(artifact_path, tar) return artifact_path
9,537
32.70318
98
py
tvm
tvm-main/python/tvm/meta_schedule/builder/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.builder package. Meta Schedule builders that translate IRModule to runtime.Module, and then export """ from .builder import Builder, BuilderInput, BuilderResult, PyBuilder, create from .local_builder import LocalBuilder
1,031
42
76
py
tvm
tvm-main/python/tvm/meta_schedule/builder/builder.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Meta Schedule builders that translate IRModule to runtime.Module, and then export""" from typing import Callable, Dict, List, Optional, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.ir import IRModule from tvm.runtime import NDArray, Object from tvm.target import Target from .. import _ffi_api @register_object("meta_schedule.BuilderInput") class BuilderInput(Object): """The builder's input. Parameters ---------- mod : IRModule The IRModule to be built. target : Target The target to be built for. params: Optional[Dict[str, NDArray]] The parameters for Relay build module """ mod: IRModule target: Target params: Optional[Dict[str, NDArray]] def __init__( self, mod: IRModule, target: Target, params: Optional[Dict[str, NDArray]] = None, ) -> None: """Constructor. Parameters ---------- mod : IRModule The IRModule to be built. target : Target The target to be built for. params: Optional[Dict[str, NDArray]] The parameters for Relay build module """ self.__init_handle_by_constructor__( _ffi_api.BuilderInput, # type: ignore # pylint: disable=no-member mod, target, params, ) @register_object("meta_schedule.BuilderResult") class BuilderResult(Object): """The builder's result. Parameters ---------- artifact_path : Optional[str] The path to the artifact. error_msg : Optional[str] The error message. """ artifact_path: Optional[str] error_msg: Optional[str] def __init__( self, artifact_path: Optional[str], error_msg: Optional[str], ) -> None: """Constructor. Parameters ---------- artifact_path : Optional[str] The path to the artifact. error_msg : Optional[str] The error message. """ self.__init_handle_by_constructor__( _ffi_api.BuilderResult, # type: ignore # pylint: disable=no-member artifact_path, error_msg, ) @register_object("meta_schedule.Builder") class Builder(Object): """The abstract builder interface.""" BuilderType = Union["Builder", Literal["local"]] def build(self, build_inputs: List[BuilderInput]) -> List[BuilderResult]: """Build the given inputs. Parameters ---------- build_inputs : List[BuilderInput] The inputs to be built. Returns ------- build_results : List[BuilderResult] The results of building the given inputs. """ return _ffi_api.BuilderBuild(self, build_inputs) # type: ignore # pylint: disable=no-member @staticmethod def create( # pylint: disable=keyword-arg-before-vararg kind: Literal["local"] = "local", *args, **kwargs, ) -> "Builder": """Create a Builder. Parameters ---------- kind : Literal["local"] The kind of the builder. For now, only "local" is supported. Returns ------- builder : Builder The builder created. """ from . import LocalBuilder # pylint: disable=import-outside-toplevel if kind == "local": return LocalBuilder(*args, **kwargs) # type: ignore raise ValueError(f"Unknown Builder: {kind}") create = Builder.create # pylint: disable=invalid-name @register_object("meta_schedule.PyBuilder") class _PyBuilder(Builder): """ A TVM object builder to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyBuilder """ def __init__(self, f_build: Callable = None): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.BuilderPyBuilder, # type: ignore # pylint: disable=no-member f_build, ) class PyBuilder: """ An abstract builder with customized build method on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = {"cls": _PyBuilder, "methods": ["build"]} def build(self, build_inputs: List[BuilderInput]) -> List[BuilderResult]: """Build the given inputs. Parameters ---------- build_inputs : List[BuilderInput] The inputs to be built. Returns ------- build_results : List[BuilderResult] The results of building the given inputs. """ raise NotImplementedError
5,664
27.325
100
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Per-block schedule rules in MetaSchedule""" from . import cpu, cuda, generic, x86
870
44.842105
62
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/generic/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Per-block schedule rules in MetaSchedule for generic cases"""
850
46.277778
64
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/cuda/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Per-block schedule rules in MetaSchedule for target key 'cuda'""" from . import layout_transform
886
43.35
68
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/cuda/layout_transform.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """layout_transform scheduling rule for cuda.""" import math from collections import deque from typing import List, Optional, Tuple, Union import tvm from tvm import meta_schedule from tvm.tir.schedule import BlockRV, ExprRV, LoopRV ## Tiling layout transforms: # Assume we have an input shape of [A, B, C, D] and want to layout transform # ABCD --> DBAC so the output shape would be [D, B, A, C]. # # Consider reading from the input buffer in a cache-friendly fashion on CPU. We would # expect a loop structure like: # lAr, lBr, lCr, lDr = T.grid(A, B, C, D) # # Meanwhile consider writing to the output buffer in a cache-friendly fashion on CPU: # lDw, lBw, lAw, lCw = T.grid(D, B, A, C) # # Clearly in many scenarios it is impossible to guarantee contiguous writes and reads # within a single loop due to non-adjacent dimensions. Instead we work on transposing some # small sub-tensor of our input writing and then reading from shared memory. We must now # construct our submatrix so that reading and writing can both be done with some contiguous # access in global memory. # # Consider the case of a 2D transpose. For example [1024, 2048] -> [2048, 1024]. # We note that if we deal with a submatrix of shape [32, 32] which corresponds # to the dimension of our input tensor, then rows of the submatrix are contiguous # in the input tensor. Meanwhile, columns of our submatrix are contiguous in our # output vector. Therefore, with this tile shape we have opportunity to read # contiguously in our input tensor and write to shared memory, and write contiguously # to our output tensor. # # The multiple dimensional case has a similar analogue. We want to allocate shared # memory per block of [`tile_size`, `tile_size`]. We want the inner most dimension # of our shared memory to correspond to contiguous reads from the input tensor and # the outer dimension to correspond to contiguous writes into the output tensor. # # In terms of the loop structure reading from the input tensor, the inner most loops # of our tile must correspond to the inner most dimensions of the input shape, # while the outer dimensions correspond to the inner most dimensions of the output shape. # To obtain an inner tile with this loop structure we factor out a contiguous `tile_size` # chunk of our loop in the shape of interest. # # An example is probably best to show this idea: # Let's say we want a layout transform of ABCD --> DCAB. With shape # [1024_a, 2_b, 32_c, 8_d] --> [8_d, 32_c, 1024_a, 2_b] # # And tile size 32. # # Then we initially have a coalesced-read loop pattern of: # T.grid(1024_a, 2_b, 32_c, 8_d) # # To obtain an inner tile of 32, we factor 4 from 32_c and 8 from 8_d: # T.grid(1024_a, 2_b, 8_c1, 1_d1, 4_c2t, 8_d2t) # T.grid(1024_a, 2_b, 8_cr, 1_dr, 32_dim1) # # To obtain an outer tile of 32, we factor from B then A to follow contiguous write # pattern: # # T.grid(64_a1, 1_b1, 8_cr, 1_dr, 16_a2t, 2_b2t, 32_dim1) # T.grid(64_ar, 1_br, 8_cr, 1_dr, 32_dim0, 32_dim1) # # Which allows us to read a tile with our wanted properties. # For writing we use the existing analysis infrastructure to generate the structure for writing. def tile_layout_transform( sch: tvm.tir.Schedule, block_read: BlockRV, block_write: BlockRV, src_layout: str, dst_layout: str, input_shape: List[int], tile_size: ExprRV, ) -> Tuple[BlockRV, BlockRV]: """ High level tiling for layout transform block. Mutates sch in place. Parameters ---------- sch: The initial schedule. We expect `block_read` and `block_write` to correspond to the blocks which reads and writes from global memory respectively. We also expect block_read's initial loops to follow block_read: The block which reads from global memory and writes to shared memory buffer. block_write: The block which writes to global memory and reads from shared memory buffer. src_layout : The src_layout, each character should appear once and also appear in dst_layout. There should be not numeric characters and refer to potentially implicit reshapes. E.g. the transform NCHW --> NCHW4c really implies NCcHW --> NCHWc. In this case src_layout should be NCcHW. dst_layout: The dst_layout. There should not be numeric characters, e.g. NCHW4c becomes NCHWc. input_shape: The input shape after applying potentially implicit reshapes. Should match the loop extants corresponding to src_layout. tile_size: The tile size of read and writes. There will be tile_size threads per block, each of which reads up to tile_size elements. Returns ------- ret: A tuple of the block that writes to global memory, and the block that reads from global memory. """ def pad_dimension_to_at_least_number(loop: LoopRV, requested_size: int): """E.g. if loop has extant of 8 but we want 10, returns size 10 loop with padding.""" left, right = sch.split(loop, [None, requested_size]) return sch.fuse(left, right) def pad_dimension_to_factor_of_tile_size( loop: LoopRV, initial_size: int, tile_size: int = tile_size ) -> Tuple[LoopRV, int]: """ Pads loop of given size until it is divisible into tile_size. If the given size of the loop is greater than tile size. Do not pad. examples: - loop_size = 5 , tile_size = 32. loop_size --> 8 - loop_size = 5 , tile_size = 36. loop_size --> 6 - loop_size = 8 , tile_size = 32. loop_size --> 8 : since 8 already divides 32. - loop_size = 33, tile_size = 32. loop_size --> 33 : since 33 > 32. Returns padded loopRV and the new size. """ if tile_size % initial_size == 0: return loop, int(initial_size) if initial_size > tile_size or initial_size == tile_size: return loop, int(initial_size) # if initial_size > tile_size return without change, factor = 1 size = initial_size while (tile_size % size) % tile_size > 0: size += 1 return pad_dimension_to_at_least_number(loop, size), int(size) def spin_out_factor( loops: List[LoopRV], loop_extants: List[int], index: int, factor_needed: int ) -> Tuple[List[LoopRV], List[int], int]: """ Factor out the requested loop's dimensions to reach the requested factor and places the requested factor as the innermost loop. Updates the schedule in-place. E.g. say we want to factors which eventually multiply to 32 (factor_needed). Say we have the index we chose is a loop with an extant of 8. E.g. loops / loop_extants = [3, 32, 6, 8], factor_needed = 32, index=3 (dim=8) - 8 divides into 32 so we just split up the loop into two loops with extants 1 and 8. - we then keep the 1-loop in place and move the new 8-loop to back of the list of loops - ending loops / loop_extants = [3, 32, 6, 1, 8], remaining_factor_needed = 32 / 8 = 4 E.g. loops / loop_extants = [3, 32, 6, 8], factor_needed=32, index=0 (dim=3) - 3 does not divide 32, so we pad until the extant divides 32, e.g. 4 - we then split up the loop into extants 1 and 4, moving the 4 to the back - ending loops / loop_extants = [1, 32, 6, 8, 4], remaining_factor_needed = 32 / 4 = 8 E.g. loops / loop_extants = [3, 32, 6, 8], factor_needed=5, index=3 (dim=8) - 8 is larger than 5 so we immediately do the splitting routine. - the 8 extant loop becomes loops with extants 2 and 5 - ending loops / loop_extants = [1, 32, 6, 2, 5], remaining_factor_needed = 5 / 5 = 1 After updating loop ordering in place, returns the new list of loops, extants, and the remaining factor needed. """ cur_loop = loops[index] cur_extant = loop_extants[index] # Pad loops to divide evenly for factors needed, and split new_loop, new_size = pad_dimension_to_factor_of_tile_size( cur_loop, cur_extant, tile_size=factor_needed ) split_factor = min(new_size, factor_needed) new_loop_split, factored_loop = sch.split(new_loop, [None, split_factor]) factor_needed = factor_needed // split_factor # update caching loops[index] = new_loop_split loops.append(factored_loop) loop_extants[index] = math.ceil(int(new_size) / int(split_factor)) loop_extants.append(split_factor) sch.reorder(*loops) return loops, loop_extants, factor_needed def factor_dim_in_order( indices: List[int], loops: List[LoopRV], cur_loop_extants: List[int], work_needed_inner_loop: int = tile_size, ) -> Tuple[List[LoopRV], List[int]]: """Factors out the loops in the order of indices until we reach needed work. Adds new loop factors to the back in reverse order of access. Returns new list of loops and their extants. """ for i in indices: loops, cur_loop_extants, work_needed_inner_loop = spin_out_factor( loops, cur_loop_extants, i, work_needed_inner_loop ) if work_needed_inner_loop == 1: break return loops, cur_loop_extants def get_high_level_loop_structure( block_read: BlockRV, input_shape: List[int], src_layout: str, dst_layout: str ): """Runs the factorization described above.""" # index 0 ... rank - 1 will always correspond to original loops # perhaps after they have been factored. rank = len(input_shape) loops = sch.get_loops(block_read) cur_loop_extants = list(input_shape) # Factor dim0 tile size and fuse things together loops, cur_loop_extants = factor_dim_in_order( list(range(rank - 1, -1, -1)), loops, cur_loop_extants, work_needed_inner_loop=tile_size, ) # The factors which multiply to tile_size are now in back of our # list of loops. However because we added them by traversing the inner # dimensions, they are actually reversed order to guarantee the best access # so reorder before fusing. loops = loops[:rank] + loops[rank:][::-1] cur_loop_extants = cur_loop_extants[:rank] + cur_loop_extants[rank::-1] sch.reorder(*loops) dim0_loop_tiled = sch.fuse(*loops[rank:]) loops = loops[:rank] loops.append(dim0_loop_tiled) cur_loop_extants = cur_loop_extants[:rank] cur_loop_extants.append(tile_size) # Same thing with dim1 # [:rank + 1], since we placed dim0_loop_tiled in the end which we want to keep loops, cur_loop_extants = factor_dim_in_order( list( ( src_layout.index(dst_layout[loop_index_dst]) for loop_index_dst in range(rank - 1, -1, -1) ) ), loops, cur_loop_extants, work_needed_inner_loop=tile_size, ) loops = loops[: rank + 1] + loops[rank + 1 :][::-1] cur_loop_extants = cur_loop_extants[: rank + 1] + cur_loop_extants[rank + 1 :: -1] sch.reorder(*loops) dim1_loop_tiled = sch.fuse(*loops[rank + 1 :]) loops = loops[: rank + 1] loops.append(dim1_loop_tiled) cur_loop_extants = cur_loop_extants[: rank + 1] cur_loop_extants.append(tile_size) # After this we have loops: [loop1, loop2, loop3 ... dim0_tiled, dim1_tiled] get_high_level_loop_structure(block_read, input_shape, src_layout, dst_layout) # If there are insufficient elements, than dim1_tiled or dim0_tiled might be too small # In all likelihood you should use a smaller tile, but I don't want things to crash. loops = sch.get_loops(block_read) loops[-1] = pad_dimension_to_at_least_number(loops[-1], tile_size) loops[-2] = pad_dimension_to_at_least_number(loops[-2], tile_size) # We want the dim0 and dim1 parent loops to be the inner most. Right now dim1 is inner-msot # and we just need to move dim0 in (last dimension of dst). # Recall right now structure is at least [l1 l2 ... ln, dim0_tiled, dim1_tiled] # where n >= 2. dim0_loop_index = src_layout.index(dst_layout[-1]) dim0_loop = loops.pop(dim0_loop_index) loops = loops[:-3] + [dim0_loop, loops[-3]] + loops[-2:] sch.reorder(*loops) # After this loops are: [outer_loop (block binding), dim0_tiled, dim1_tiled] outer_loop = sch.fuse(*loops[:-2]) # Now that we have the high level loop structure, we can use reverse_compute_at magic # To get the proper loop structure for writing! This is also as coalesced as possible # already. sch.reverse_compute_at(block_write, outer_loop) # Fuse all inner loops for the write into 2 loops, grab inner loops for both read # and write block which have locality (we will bind these to threadIdx) fused_write_loop = sch.fuse(*sch.get_loops(block_write)[1:]) _, inner_write_loop = sch.split(fused_write_loop, [None, tile_size]) inner_read_loop = sch.get_loops(block_read)[-2] sch.bind(loop=outer_loop, thread_axis="blockIdx.x") sch.bind(loop=inner_write_loop, thread_axis="threadIdx.x") sch.bind(loop=inner_read_loop, thread_axis="threadIdx.x") return block_write, block_read def create_cached_read( sch: tvm.tir.Schedule, block_write: BlockRV, orig_input_shape: List[int], orig_src_layout: str, orig_dst_layout: str, ) -> Tuple[BlockRV, List[int], str, str]: """ Creates the cached read block with expected structure. Loop extants should follow the input shape closely. E.g. if the input is [2, 6, 8], we expect our loop structure to be T.grid(2, 6, 8). Possibly reshape to handle implicit reshapes, in which case we will match the implicit reshape shape. Layout transform allows semantics like NCHW --> NCHW4c. Which involves splitting the original C axis into contiguous 4-element chunks. This axis is then moved to the end (NCHWc). This is guaranteed by the operator to be done without additional padding. To handle this we just split the associating axis (prev. type checking ensures C is divisible by 4)in src_layout found in block_read. E.g. NCHW -> NCHW4c now becomes NC4cHW -> NCHW4c. Note: NCHW4c --> NCHW is not allowed, so the only numeric digits will be in dst. The returned layout strings will be santized and made compatible. E.g. NCHW --> NCHW4c becomes NCcHW --> NCHWc. TODO(AndrewZhaoLuo): Investigate using proper memory alignment to avoid bank conflict. Parameters ---------- sch: The initial schedule. We expect `block_read`. We also expect block_read's initial loops to follow the original input shape. block_read: The block which reads from global memory and writes to shared memory buffer. orig_input_shape: The input shape of the input buffer to the primfunc. orig_src_layout: The original src_layout string. orig_dst_layout: The original dst_layout string. Returns ------- ret: A tuple of the cached read block, new input shape of shared memory buffer, the new src_layout, and new dst_layout string. """ # Figure out split dimensions, entries are (loop index in src_layout, split amount) split_dimensions: List[Tuple[int, int]] = [] # This is without numeric digits, e.g. NCHW4c -> NCHWc new_dst_layout = [] # Use state machine to parse NCHW4c string split_size = 0 for char in orig_dst_layout: if char.isnumeric(): split_size = split_size * 10 + int(char) else: if char.islower(): # hit axis like 'c', need to find parent axis 'C' in src_layout src_layout_index = orig_src_layout.index(char.upper()) split_dimensions.append((src_layout_index, split_size)) split_size = 0 new_dst_layout.append(char) # If no splits were detected we are done if len(split_dimensions) == 0: block_read = sch.cache_read(block_write, 0, "shared") return block_read, orig_input_shape, orig_src_layout, orig_dst_layout # Calculate final input shapes, each of these are a single element for unsplit dims # and tuples for split dims associated with the two new axis input_shape: List[Union[int, Tuple]] = list(orig_input_shape) new_src_layout: List[Union[str, Tuple]] = list(orig_src_layout) for src_layout_split_index, split_factor in split_dimensions: dimension_name = orig_src_layout[src_layout_split_index] new_src_layout[src_layout_split_index] = (dimension_name, dimension_name.lower()) input_shape[src_layout_split_index] = ( orig_input_shape[src_layout_split_index] // split_factor, split_factor, ) # Unpack any tuples introduced via appending def unpack_list(target_list) -> List: output: List = [] for ele in target_list: if isinstance(ele, tuple): output.extend(ele) else: output.append(ele) return output new_src_layout_str = "".join(unpack_list(new_src_layout)) new_dst_layout_str = "".join(unpack_list(new_dst_layout)) # Write block loop extants match dst_to_src_map = [new_dst_layout_str.index(dim) for dim in new_src_layout_str] block_read = sch.reindex_cache_read( block_write, read_buffer_index=0, index_map=tvm.tir.IndexMap.from_func( lambda *loops: [loops[dst_to_src_map[i]] for i, _ in enumerate(loops)], ndim=len(new_src_layout_str), ), storage_scope="shared", ) loops_read = sch.get_loops(block_read) sch.reorder( *[loops_read[new_dst_layout_str.index(dst_dim_name)] for dst_dim_name in new_src_layout_str] ) return block_read, unpack_list(input_shape), new_src_layout_str, new_dst_layout_str def auto_inline_into(sch: tvm.tir.Schedule, start_block: BlockRV) -> BlockRV: """ Inlines given start_block's consumers and future dependencies into start_block. Parameters ---------- sch: The initial schedule. start_block: The block to inline into, should be a block which reads and writes to global memory, doing layout transform. Returns ------- ret: The new block inlined into it's consumers. """ # Rules defined by DefaultCUDA schedule_rule set. autoinline_rule = meta_schedule.schedule_rule.AutoInline( into_producer=True, into_consumer=False, inline_const_tensor=True, disallow_if_then_else=False, require_injective=False, require_ordered=False, ) fringe = deque(sch.get_consumers(start_block)) visited = set() while len(fringe) > 0: cur_block = fringe.popleft() if cur_block in visited: continue visited.add(cur_block) consumer_blocks = sch.get_consumers(cur_block) fringe.extend(consumer_blocks) sch = autoinline_rule.apply(sch, cur_block)[0] def get_max_tile_size() -> int: """Returns the max tile size. This is assuming only threads in a warp can have coalesced accesses. 32 is the default if no target information can be gotten. """ max_tile_size = 32 cur_target = tvm.target.Target.current() if cur_target is not None and hasattr(cur_target, "thread_warp_size"): max_tile_size = int(cur_target.thread_warp_size) return max_tile_size @tvm.register_func("meta_schedule.cuda.layout_transform") def cuda_layout_transform_schedule_rule( sch: tvm.tir.Schedule, block: BlockRV, testing_tile_sizes: Optional[List[int]] = None ) -> List[tvm.tir.Schedule]: """ Applies tiling scheme to layout transform task (potentially fused with other injective funcs). Returned schedules will be the default schedule, as well as tiled versions with tile_size in the range of 2,3...threads_per_warp. This is assuming only threads in a warp can have coalesced accesses. 32 is the default if no target information can be gotten. Parameters ---------- sch: The initial schedule. block: The block corresponding to the layout transform. Should be a block which reads and writes to global memory, doing layout transform. testing_tile_sizes: A list of tile sizes to try, overriding normal settings. For testing. None means ignore. Else overrides normal settings of tile sizes to try. Returns ------- ret: A list of new schedules to try. """ # Info needed for tiling src_layout = sch.get_sref(block).stmt.annotations["src_layout"] dst_layout = sch.get_sref(block).stmt.annotations["dst_layout"] input_shape = [int(c) for c in sch.get_sref(block).stmt.annotations["input_shape"]] schedules = [] # Always include the default schedules which will be handled via AutoBind schedule rule # Except during testing if not testing_tile_sizes: schedules.append(sch) sch = sch.copy() # Inline consumers of the layout transform into the layout transform block. # Normally default for injective schedules but must manually be called in new schedule rule # for consumers of the layout transform. TODO(AndrewZhaoLuo): Figure out why this is the case. auto_inline_into(sch, block) # Setup up basic structure of schedule of creating read into shared mem, before applying tiling # Outer loop structure of read block matches that of src_layout # E.g. if input_shape is [4, 6, 8]. Loops for read block will be # for i, j, k in T.grid(4, 6, 8): # ... # Read block will read from global memory coalesced at the start # Assume write to output global memory is coalesced in block_write # # This also handles the case where there is an implicit reshape going on. # e.g. NCHW -> NCHW4c which is equivalent to reshaping NCHW # to NCcHW and then applying the new layout where the extant of c is 4. # Grab final input shape and src and dst layouts with possible implicit reshape. block_read, input_shape, src_layout, dst_layout = create_cached_read( sch, block, input_shape, src_layout, dst_layout ) # Try tile size 2,3...threads_per_warp as tile size of 1 has no coaslescing. if testing_tile_sizes is None: tile_sizes = list(range(2, get_max_tile_size() + 1)) else: tile_sizes = testing_tile_sizes for tile_size in tile_sizes: new_sch = sch.copy() tile_layout_transform( new_sch, block_read, block, src_layout, dst_layout, input_shape, tile_size ) schedules.append(new_sch) return schedules
23,824
39.796233
100
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/x86/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Per-block schedule rules in MetaSchedule for target key 'x86'"""
853
46.444444
67
py
tvm
tvm-main/python/tvm/meta_schedule/schedule/cpu/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Per-block schedule rules in MetaSchedule for target key 'cpu'"""
853
46.444444
67
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/parallel_vectorize_unroll.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Rule that mark parallelize, vectorize and unroll to the root block. The mark will be applied to each block in a follow-up post processor""" from typing import List, Optional from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.ParallelizeVectorizeUnroll") class ParallelizeVectorizeUnroll(ScheduleRule): """Rule that mark parallelize, vectorize and unroll to the root block. The mark will be applied to each block in a follow-up post processor Parameters ---------- max_jobs_per_core: int The maximum number of jobs to be launched per CPU core. It sets the upper limit of CPU parallelism, i.e. `num_cores * max_jobs_per_core`. Use -1 to disable parallelism. max_vectorize_extent: int The maximum extent to be vectorized. It sets the upper limit of the hardware target vectorization. Use -1 to disable vectorization. unroll_max_steps: Optional[List[int]] The options of the maximum number of unroll steps to be done. Use None to disable unroll unroll_explicit: bool Whether to explicitly unroll the loop, or just add an "unroll" pragma """ def __init__( self, max_jobs_per_core: int = 16, max_vectorize_extent: int = 16, unroll_max_steps: Optional[List[int]] = None, unroll_explicit: bool = True, ) -> None: if unroll_max_steps is None: unroll_max_steps = [] self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleParallelizeVectorizeUnroll, # type: ignore # pylint: disable=no-member max_jobs_per_core, max_vectorize_extent, unroll_max_steps, unroll_explicit, )
2,592
38.892308
104
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/random_compute_location.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Rule that randomly select a compute-at location for a free block""" from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.RandomComputeLocation") class RandomComputeLocation(ScheduleRule): """A rule that randomly select a compute-at location for a free block""" def __init__(self) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleRandomComputeLocation, # type: ignore # pylint: disable=no-member )
1,324
40.40625
99
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/add_rfactor.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Add-rfactor Rule that add-rfactor to some blocks if needed""" from typing import Optional from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.AddRFactor") class AddRFactor(ScheduleRule): """Rules for add-rfactor to some blocks if needed. Parameters ---------- max_jobs_per_core: int The maximum number of jobs to be launched per CPU core. It sets the uplimit of CPU parallelism, i.e. `num_cores * max_jobs_per_core`. Use -1 to disable parallelism. max_innermost_factor: Optional[int] = None The maximum size of the innermost factor. None means no limit. """ def __init__( self, max_jobs_per_core: int = 16, max_innermost_factor: Optional[int] = None, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleAddRFactor, # type: ignore # pylint: disable=no-member max_jobs_per_core, max_innermost_factor, )
1,834
35.7
90
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/schedule_rule.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ Meta Schedule schedule rules are used for modification of blocks in a schedule. See also PostOrderApply. """ from typing import TYPE_CHECKING, Callable, List # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from tvm.tir.schedule import BlockRV, Schedule from .. import _ffi_api from ..utils import _get_default_str if TYPE_CHECKING: from ..tune_context import TuneContext @register_object("meta_schedule.ScheduleRule") class ScheduleRule(Object): """Rules to modify a block in a schedule.""" def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the schedule rule with a tune context. Parameters ---------- context : TuneContext The tuning context for initializing the schedule rule. """ _ffi_api.ScheduleRuleInitializeWithTuneContext( # type: ignore # pylint: disable=no-member self, context ) def apply(self, sch: Schedule, block: BlockRV) -> List[Schedule]: """Apply a schedule rule to the specific block in the given schedule. Parameters ---------- sch : tvm.tir.Schedule The schedule to be modified. block : BlockRV The specific block to apply the schedule rule. Returns ------- design_spaces : List[tvm.tir.Schedule] The list of schedules generated by applying the schedule rule. """ return _ffi_api.ScheduleRuleApply( # type: ignore # pylint: disable=no-member self, sch, block ) def clone(self) -> "ScheduleRule": """Deep clone the schedule rule. Returns ------- cloned_rule : ScheduleRule The cloned schedule rule. """ return _ffi_api.ScheduleRuleClone(self) # type: ignore # pylint: disable=no-member @staticmethod def create(kind: Literal["llvm", "cuda", "cuda-tensorcore", "hexagon"]) -> List["ScheduleRule"]: """Create a list of schedule rules for the given kind. Parameters ---------- kind : Literal["llvm", "cuda", "cuda-tensorcore", "hexagon"] The kind of the schedule rules. Returns ------- rules : List[ScheduleRule] The list of schedule rules. """ funcs = { # pylint: disable=no-member "llvm": _ffi_api.ScheduleRuleDefaultLLVM, # type: ignore "cuda": _ffi_api.ScheduleRuleDefaultCUDA, # type: ignore "cuda-tensorcore": _ffi_api.ScheduleRuleDefaultCUDATensorCore, # type: ignore "hexagon": _ffi_api.ScheduleRuleDefaultHexagon, # type: ignore # pylint: enable=no-member } for k, v in funcs.items(): if k == kind: return v() raise ValueError(f"Unsupported kind {kind} for schedule rule creation.") create = ScheduleRule.create # pylint: disable=invalid-name @register_object("meta_schedule.PyScheduleRule") class _PyScheduleRule(ScheduleRule): """ A TVM object schedule rule to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyScheduleRule """ def __init__( self, f_initialize_with_tune_context: Callable = None, f_apply: Callable = None, f_clone: Callable = None, f_as_string: Callable = None, ): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.ScheduleRulePyScheduleRule, # type: ignore # pylint: disable=no-member f_initialize_with_tune_context, f_apply, f_clone, f_as_string, ) class PyScheduleRule: """ An abstract schedule rule with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyScheduleRule, "methods": ["_initialize_with_tune_context", "apply", "clone", "__str__"], } def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the schedule rule with a tune context. Parameters ---------- context : TuneContext The tuning context for initializing the schedule rule. """ raise NotImplementedError def apply(self, sch: Schedule, block: BlockRV) -> List[Schedule]: """Apply a schedule rule to the specific block in the given schedule. Parameters ---------- sch : Schedule The schedule to be modified. block : BlockRV The specific block to apply the schedule rule. Returns ------- design_spaces : List[Schedule] The list of schedules generated by applying the schedule rule. """ raise NotImplementedError def clone(self) -> ScheduleRule: """Deep clone the schedule rule. Returns ------- cloned_rule : ScheduleRule The cloned schedule rule. """ raise NotImplementedError def __str__(self) -> str: """Get the schedule rule as string with name. Return ------ result : str Get the schedule rule as string with name. """ return _get_default_str(self)
6,338
30.537313
100
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/auto_bind.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Auto-bind Rule that binds blocks to threads if needed""" from typing import List, Optional from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.AutoBind") class AutoBind(ScheduleRule): """Auto bind loops around the block to BlockIdx and ThreadIdx Parameters ---------- max_threadblocks: int The maximum number of threadblock on GPU. thread_extents: Optional[List[int]] Candidates of thread axis extent. max_threads_per_block: int The maximum number of threads per block, if it is known when this schedule rule is created. """ def __init__( self, max_threadblocks: int = 256, thread_extents: Optional[List[int]] = None, max_threads_per_block: int = -1, ) -> None: if thread_extents is None: thread_extents = [32, 64, 128, 256, 512, 1024] self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleAutoBind, # type: ignore # pylint: disable=no-member max_threadblocks, thread_extents, max_threads_per_block, )
1,958
35.277778
99
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/auto_inline.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Auto-Inline. Rule that inlines spatial blocks if it satisfies some conditions""" from typing import List, Optional from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.AutoInline") class AutoInline(ScheduleRule): """Rule that inlines spatial blocks if it satisfies some conditions Parameters ---------- into_producer : bool If allows to inline a block into its producer into_consumer : bool If allows to inline a block into its consumer inline_const_tensor : bool Always inline constant tensors disallow_if_then_else : bool Always disallow if-then-else-like constructs require_injective : bool Always require the read-to-write mapping to be ordered require_ordered : bool Always require the read-to-write mapping to be injective disallow_op : Optional[List[str]] The operators that are disallowed in auto inline """ def __init__( self, into_producer: bool, into_consumer: bool, inline_const_tensor: bool, disallow_if_then_else: bool, require_injective: bool, require_ordered: bool, disallow_op: Optional[List[str]] = None, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleAutoInline, # type: ignore # pylint: disable=no-member into_producer, into_consumer, inline_const_tensor, disallow_if_then_else, require_injective, require_ordered, disallow_op, ) @register_object("meta_schedule.InlineConstantScalars") class InlineConstantScalars(ScheduleRule): """Inline blocks that produce a constant scalar. Such blocks get in the way of ReverseComputeInline during AutoInline, since they are also counted as a producer block unless they are inlined first. So it is recommended to run InlineConstantScalars before AutoInline. """ def __init__( self, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleInlineConstantScalars, # type: ignore # pylint: disable=no-member )
3,027
34.623529
99
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/multi_level_tiling.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Multi-level tiling with reuse.""" from typing import Any, Dict, List, Mapping, NamedTuple, Optional, Callable from tvm.tir.schedule import Schedule, BlockRV from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule class ReuseType(NamedTuple): """Reuse type.""" req: str levels: List[int] scope: str def as_dict(self) -> Dict[str, Any]: """Return the dict representation of the reuse type.""" return { "req": self.req, "levels": self.levels, "scope": self.scope, } @register_object("meta_schedule.MultiLevelTiling") class MultiLevelTiling(ScheduleRule): """Multi-level tiling with reuse. Parameters ---------- structure : str The tiling structure. Recommended: - 'SSRSRS' on CPU - 'SSSRRSRS' on GPU tile_bind : Optional[List[str]] For each level of tiles, which thread axis it is bound to. Recommended: - None on CPU - [blockIdx.x, vthread.x, threadIdx.x] on GPU max_innermost_factor : Optional[int] The maximum size of the innermost factor. None means no limit vector_load_lens : Optional[List[int]] The length of vector lane in vectorized cooperative fetching. None means disable vectorization reuse_read : Optional[ReuseType] Data reuse configuration for reading. None means no reuse. reuse_write : Optional[ReuseType] Data reuse configuration for writing. None means no reuse. filter_fn: Optional[Callable[[Schedule, BlockRV], bool]] A function that can be passed to overwrite the default condition for applying MultiLevelTiling to a block. This is useful if there is a need to apply MultiLevelTiling to an operation / block which is ignored by default. This function should return True for a block that should be tiled (based on the block name, for example). """ def __init__( self, structure: str, tile_binds: Optional[List[str]] = None, max_innermost_factor: Optional[int] = None, vector_load_lens: Optional[List[int]] = None, reuse_read: Optional[ReuseType] = None, reuse_write: Optional[ReuseType] = None, filter_fn: Optional[Callable[[Schedule, BlockRV], bool]] = None, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleMultiLevelTiling, # type: ignore # pylint: disable=no-member structure, tile_binds, max_innermost_factor, vector_load_lens, reuse_read.as_dict() if reuse_read is not None else None, reuse_write.as_dict() if reuse_write is not None else None, filter_fn, ) @register_object("meta_schedule.MultiLevelTilingWithIntrin") class MultiLevelTilingWithIntrin(ScheduleRule): """Extension of MultiLevelTiling for auto-tensorizing with a single intrinsic. Parameters ---------- intrin_name : str The name of a tensor intrinsic, must be registerd via TensorIntrin.register(...) beforehand structure : str The tiling structure. Recommended: - 'SSRSRS' on CPU - 'SSSRRSRS' on GPU tile_bind : Optional[List[str]] For each level of tiles, which thread axis it is bound to. Recommended: - None on CPU - [blockIdx.x, vthread.x, threadIdx.x] on GPU max_innermost_factor : Optional[int] The maximum size of the innermost factor. None means no limit vector_load_lens : Optional[List[int]] The length of vector lane in vectorized cooperative fetching. None means disable vectorization reuse_read : Optional[ReuseType] Data reuse configuration for reading. None means no reuse. reuse_write : Optional[ReuseType] Data reuse configuration for writing. None means no reuse. """ def __init__( self, intrin_name: str, structure: str, tile_binds: Optional[List[str]] = None, max_innermost_factor: Optional[int] = None, vector_load_lens: Optional[List[int]] = None, reuse_read: Optional[ReuseType] = None, reuse_write: Optional[ReuseType] = None, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleMultiLevelTilingWithIntrin, # type: ignore # pylint: disable=no-member intrin_name, structure, tile_binds, max_innermost_factor, vector_load_lens, reuse_read.as_dict() if reuse_read is not None else None, reuse_write.as_dict() if reuse_write is not None else None, ) @register_object("meta_schedule.MultiLevelTilingTensorCore") class MultiLevelTilingTensorCore(ScheduleRule): """Extension of MultiLevelTiling for auto-tensorizing with multiple groups of candidate tensor core intrinsics. Parameters ---------- intrin_groups : List[Mapping[str, str]] A list of groups of tensor core intrinsics. The map should contains key "init", "load_a", "load_b", "compute", "store", which represent the tensor intrin for initialization, loading operand A, loading operand B, tensor core computation, storing the result. The value of the map should be names of tensor intrinsics, must be registerd via TensorIntrin.register(...) beforehand structure : str The tiling structure. Recommended: - 'SSSRRSRS' on GPU tile_bind : Optional[List[str]] For each level of tiles, which thread axis it is bound to. Recommended: - [blockIdx.y, vthread.x, threadIdx.y] on GPU max_innermost_factor : Optional[int] The maximum size of the innermost factor. None means no limit vector_load_lens : Optional[List[int]] The length of vector lane in vectorized cooperative fetching. None means disable vectorization reuse_read : Optional[ReuseType] Data reuse configuration for reading. None means no reuse. reuse_write : Optional[ReuseType] Data reuse configuration for writing. None means no reuse. use_software_pipeline : bool Whether to use the software pipeline. """ def __init__( self, intrin_groups: List[Mapping[str, str]], structure: str, tile_binds: Optional[List[str]] = None, max_innermost_factor: Optional[int] = None, vector_load_lens: Optional[List[int]] = None, reuse_read: Optional[ReuseType] = None, reuse_write: Optional[ReuseType] = None, use_software_pipeline: bool = False, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleMultiLevelTilingTensorCore, # type: ignore # pylint: disable=no-member intrin_groups, structure, tile_binds, max_innermost_factor, vector_load_lens, reuse_read.as_dict() if reuse_read is not None else None, reuse_write.as_dict() if reuse_write is not None else None, use_software_pipeline, ) @register_object("meta_schedule.MultiLevelTilingWideVector") class MultiLevelTilingWideVector(ScheduleRule): """Extension of MultiLevelTiling for backends with wide vectors. The loop over the innermost spatial axis of the output buffer is always vectorized with the maximum vector length. Parameters ---------- structure : str The tiling structure. 'SSRSRS' is recommended. vector_length_in_bits: int The length of a vector register in bits. max_innermost_factor : Optional[int] The maximum size of the innermost factor. None means no limit reuse_read : Optional[ReuseType] Data reuse configuration for reading. None means no reuse. reuse_write : Optional[ReuseType] Data reuse configuration for writing. None means no reuse. """ def __init__( self, structure: str, vector_length_in_bits: int, max_innermost_factor: Optional[int] = None, reuse_read: Optional[ReuseType] = None, reuse_write: Optional[ReuseType] = None, ) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleMultiLevelTilingWideVector, # type: ignore # pylint: disable=no-member structure, vector_length_in_bits, max_innermost_factor, reuse_read.as_dict() if reuse_read is not None else None, reuse_write.as_dict() if reuse_write is not None else None, )
9,398
38.995745
104
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/apply_custom_rule.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Create a rule that applies customized rules registered using block attribute `schedule_rule`. The rule will be dispatched according to target keys.""" from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.ApplyCustomRule") class ApplyCustomRule(ScheduleRule): """A rule that applies customized rules registered using block attribute `schedule_rule`. The rule will be dispatched according to target keys.""" def __init__(self) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleApplyCustomRule, # type: ignore # pylint: disable=no-member )
1,467
42.176471
96
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.schedule_rule package. Meta Schedule schedule rules are used for modification of blocks in a schedule. See also PostOrderApply. """ from .add_rfactor import AddRFactor from .apply_custom_rule import ApplyCustomRule from .auto_bind import AutoBind from .auto_inline import AutoInline, InlineConstantScalars from .cross_thread_reduction import CrossThreadReduction from .multi_level_tiling import ( MultiLevelTiling, MultiLevelTilingTensorCore, MultiLevelTilingWideVector, MultiLevelTilingWithIntrin, ReuseType, ) from .parallel_vectorize_unroll import ParallelizeVectorizeUnroll from .random_compute_location import RandomComputeLocation from .schedule_rule import PyScheduleRule, ScheduleRule
1,524
40.216216
65
py
tvm
tvm-main/python/tvm/meta_schedule/schedule_rule/cross_thread_reduction.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Rules which apply cross-thread reduction to some reduction blocks correspondingly when needed""" from typing import List from tvm._ffi import register_object from .. import _ffi_api from .schedule_rule import ScheduleRule @register_object("meta_schedule.CrossThreadReduction") class CrossThreadReduction(ScheduleRule): """A schedule rule which applies cross-thread reduction to some reduction blocks correspondingly when needed Parameters ---------- thread_extents: List[int] Candidates of thread axis extent (values are required to be positive). """ def __init__(self, thread_extents: List[int]) -> None: self.__init_handle_by_constructor__( _ffi_api.ScheduleRuleCrossThreadReduction, # type: ignore # pylint: disable=no-member thread_extents, )
1,618
37.547619
99
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutator.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Meta Schedule Mutator.""" from typing import TYPE_CHECKING, Callable, Dict, Optional # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from tvm.tir.schedule import Trace from .. import _ffi_api from ..utils import _get_default_str if TYPE_CHECKING: from ..tune_context import TuneContext class Mutator(Object): """Mutator is designed to mutate the trace to explore the design space.""" def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the mutator with a tune context. Parameters ---------- context : TuneContext The tuning context for initializing the mutator. """ _ffi_api.MutatorInitializeWithTuneContext( # type: ignore # pylint: disable=no-member self, context ) def apply(self, trace: Trace) -> Optional[Trace]: """Apply the mutator function to the given trace. Parameters ---------- trace : Trace The given trace for mutation. Returns ------- trace : Optional[Trace] None if mutator failed, otherwise return the mutated trace. """ return _ffi_api.MutatorApply(self, trace, -1) # type: ignore # pylint: disable=no-member def clone(self) -> "Mutator": """Clone the mutator. Returns ------- mutator : Mutator The cloned mutator. """ return _ffi_api.MutatorClone(self) # type: ignore # pylint: disable=no-member @staticmethod def create( kind: Literal[ "llvm", "cuda", "cuda-tensorcore", "hexagon", ] ) -> Dict["Mutator", float]: """Create a list of default mutators. Parameters ---------- kind : Literal["llvm", "cuda", "cuda-tensorcore", "hexagon"] The kind of mutators. Returns ------- mutators : List[Mutator] The list of mutators. """ funcs = { # pylint: disable=no-member "llvm": _ffi_api.MutatorDefaultLLVM, # type: ignore "cuda": _ffi_api.MutatorDefaultCUDA, # type: ignore "cuda-tensorcore": _ffi_api.MutatorDefaultCUDATensorCore, # type: ignore "hexagon": _ffi_api.MutatorDefaultHexagon, # type: ignore # pylint: enable=no-member } for k, v in funcs.items(): if k == kind: return v() raise ValueError(f"Unsupported kind {kind} for mutator creation.") create = Mutator.create # pylint: disable=invalid-name @register_object("meta_schedule.PyMutator") class _PyMutator(Mutator): """ A TVM object mutator to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyMutator """ def __init__( self, f_initialize_with_tune_context: Callable = None, f_apply: Callable = None, f_clone: Callable = None, f_as_string: Callable = None, ): """Constructor.""" self.__init_handle_by_constructor__( _ffi_api.MutatorPyMutator, # type: ignore # pylint: disable=no-member f_initialize_with_tune_context, f_apply, f_clone, f_as_string, ) class PyMutator: """ An abstract mutator with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyMutator, "methods": ["_initialize_with_tune_context", "apply", "clone", "__str__"], } def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the mutator with a tune context. Parameters ---------- context : TuneContext The tuning context for initializing the mutator. """ raise NotImplementedError def apply(self, trace: Trace, _) -> Optional[Trace]: """Apply the mutator function to the given trace. Parameters ---------- trace : Trace The given trace for mutation. Returns ------- trace : Optional[Trace] None if mutator failed, otherwise return the mutated trace. """ raise NotImplementedError def clone(self) -> Mutator: """Clone the mutator. Returns ------- mutator : Mutator The cloned mutator. """ raise NotImplementedError def __str__(self) -> str: """Get the mutator as string with name. Return ------ result : str Get the mutator as string with name. """ return _get_default_str(self)
5,779
28.191919
97
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutate_tile_size.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Mutator that mutates the decision of instruction Sample-Perfect-Tile""" from tvm._ffi.registry import register_object from .. import _ffi_api from .mutator import Mutator @register_object("meta_schedule.MutateTileSize") class MutateTileSize(Mutator): """Mutator that mutates the decision of instruction Sample-Perfect-Tile""" def __init__(self) -> None: self.__init_handle_by_constructor__( _ffi_api.MutatorMutateTileSize, # type: ignore # pylint: disable=no-member )
1,297
39.5625
87
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutate_unroll.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Mutator that mutates auto unroll step""" from tvm._ffi.registry import register_object from .. import _ffi_api from .mutator import Mutator @register_object("meta_schedule.MutateUnroll") class MutateUnroll(Mutator): """Mutator that mutates auto unroll step""" def __init__(self) -> None: self.__init_handle_by_constructor__( _ffi_api.MutatorMutateUnroll, # type: ignore # pylint: disable=no-member )
1,229
37.4375
85
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutate_parallel.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Mutator that mutates the parallel extent""" from tvm._ffi.registry import register_object from .. import _ffi_api from .mutator import Mutator @register_object("meta_schedule.MutateParallel") class MutateParallel(Mutator): """Mutator that mutates the parallel extent""" def __init__(self, max_jobs_per_core: int) -> None: """Mutator that mutates the parallel extent""" self.__init_handle_by_constructor__( _ffi_api.MutatorMutateParallel, # type: ignore # pylint: disable=no-member max_jobs_per_core, )
1,351
38.764706
87
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutate_thread_binding.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Mutator that mutates the thread binding extent""" from tvm._ffi.registry import register_object from .. import _ffi_api from .mutator import Mutator @register_object("meta_schedule.MutateThreadBinding") class MutateThreadBinding(Mutator): """Mutator that mutates the binding extent""" def __init__(self) -> None: """Mutator that mutates the binding extent""" self.__init_handle_by_constructor__( _ffi_api.MutateThreadBinding, # type: ignore # pylint: disable=no-member )
1,308
38.666667
85
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.mutator package. Meta Schedule mutator that mutates the trace to explore the design space. """ from .mutator import Mutator, PyMutator from .mutate_compute_location import MutateComputeLocation from .mutate_tile_size import MutateTileSize from .mutate_thread_binding import MutateThreadBinding from .mutate_parallel import MutateParallel from .mutate_unroll import MutateUnroll
1,189
41.5
62
py
tvm
tvm-main/python/tvm/meta_schedule/mutator/mutate_compute_location.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """A mutator that mutates the compute-at location decision of SampleComputeLocation""" from tvm._ffi.registry import register_object from .. import _ffi_api from .mutator import Mutator @register_object("meta_schedule.MutateComputeLocation") class MutateComputeLocation(Mutator): """A mutator that mutates the compute-at location decision of SampleComputeLocation""" def __init__(self) -> None: self.__init_handle_by_constructor__( _ffi_api.MutatorMutateComputeLocation, # type: ignore # pylint: disable=no-member )
1,342
40.96875
94
py
tvm
tvm-main/python/tvm/meta_schedule/runner/utils.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Runner utility functions""" import itertools from typing import Any, Callable, Dict, List from ...runtime import Device, Module, ndarray from .config import EvaluatorConfig T_ARG_INFO_JSON_OBJ = List[Any] # pylint: disable=invalid-name T_ARG_INFO_JSON_OBJ_LIST = List[T_ARG_INFO_JSON_OBJ] # pylint: disable=invalid-name T_ARGUMENT = Any # pylint: disable=invalid-name T_ARGUMENT_LIST = List[T_ARGUMENT] # pylint: disable=invalid-name def alloc_argument_common( f_random_fill: Callable, device: Device, args_info: T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[T_ARGUMENT_LIST]: """Common function to allocate the arguments Parameters ---------- f_random_fill: Callable The callable function for random fill device: Device The device to allocate the arguments args_info: T_ARG_INFO_JSON_OBJ_LIST The arguments info alloc_repeat: int The number of times to repeat the allocation Returns ------- repeated_args: List[T_ARGUMENT_LIST] The allocation args """ def alloc_tensor(_, dtype, shape) -> ndarray.NDArray: arg = ndarray.empty(shape=shape, dtype=dtype, device=device) f_random_fill(arg) return arg def alloc_fail(*arg_info) -> None: raise NotImplementedError(arg_info) dispatcher: Dict[Any, Callable] = { "TENSOR": alloc_tensor, None: alloc_fail, } repeated_args: List[T_ARGUMENT_LIST] = [] for _ in range(alloc_repeat): args: T_ARGUMENT_LIST = [] arg_info: T_ARG_INFO_JSON_OBJ for arg_info in args_info: arg_type = arg_info[0] arg: Any = dispatcher.get(arg_type, None)(*arg_info) args.append(arg) repeated_args.append(args) return repeated_args def run_evaluator_common( rt_mod: Module, device: Device, evaluator_config: EvaluatorConfig, repeated_args: List[T_ARGUMENT_LIST], ) -> List[float]: """Common function to run the evaluator Parameters ---------- rt_mod: Module The runtime module device: Device The device to run the evaluator evaluator_config: EvaluatorConfig The evaluator config repeated_args: List[T_ARGUMENT_LIST] The repeated arguments Returns ------- costs: List[float] The evaluator results """ evaluator = rt_mod.time_evaluator( func_name=rt_mod.entry_name, dev=device, number=evaluator_config.number, repeat=evaluator_config.repeat, min_repeat_ms=evaluator_config.min_repeat_ms, f_preproc="cache_flush_cpu_non_first_arg" if evaluator_config.enable_cpu_cache_flush else "", ) repeated_costs: List[List[float]] = [] for args in repeated_args: device.sync() profile_result = evaluator(*args) repeated_costs.append(profile_result.results) costs = [float(cost) for cost in itertools.chain.from_iterable(repeated_costs)] return costs
3,820
30.578512
84
py
tvm
tvm-main/python/tvm/meta_schedule/runner/config.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Configurations for measurements in the runner""" import os from threading import Thread from typing import NamedTuple, Optional, Union from tvm import rpc class EvaluatorConfig(NamedTuple): """Config Details of Evaluator Parameters ---------- number: int The number of times to run this function for taking average. We call these runs as one `repeat` of measurement. repeat: int The number of times to repeat the measurement. In total, the function will be invoked (1 + number x repeat) times, where the first one is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms: int Minimum repeat time in ms. if the execution latency is too short, increase the number of runs to the given time (in ms) to reduce the measurement error. enable_cpu_cache_flush: bool Whether to flush the cache on CPU. Note ---- The total number of actual executions is 1+number*repeat because we would warm up 1 time before actual run. The number of runs would be increased if run time is below min_repeat_ms. """ number: int = 3 repeat: int = 1 min_repeat_ms: int = 100 enable_cpu_cache_flush: bool = False @staticmethod def _normalized(config: Optional["EvaluatorConfig"]) -> "EvaluatorConfig": if config is None: return EvaluatorConfig() config = EvaluatorConfig( number=config.number, repeat=config.repeat, min_repeat_ms=config.min_repeat_ms, enable_cpu_cache_flush=config.enable_cpu_cache_flush, ) return config class RPCConfig(NamedTuple): """RPC configuration Parameters ---------- tracker_host: str Host of the RPC Tracker tracker_port: int Port of the RPC Tracker tracker_key: str Key of the Tracker session_timeout_sec: float Timeout of the RPC session session_priority: int Priority of the RPC session """ tracker_host: Optional[str] = None tracker_port: Union[None, int, str] = None tracker_key: Optional[str] = None session_priority: int = 1 session_timeout_sec: int = 10 def _sanity_check(self) -> None: err_str = ( "RPCConfig.{0} is not provided. Please provide it explicitly," "or set environment variable {1}" ) if self.tracker_host is None: raise ValueError(err_str.format("tracker_host", "TVM_TRACKER_HOST")) if self.tracker_port is None: raise ValueError(err_str.format("tracker_port", "TVM_TRACKER_PORT")) if self.tracker_key is None: raise ValueError(err_str.format("tracker_key", "TVM_TRACKER_KEY")) @staticmethod def _normalized(config: Optional["RPCConfig"]) -> "RPCConfig": if config is None: config = RPCConfig() tracker_host = config.tracker_host or os.environ.get("TVM_TRACKER_HOST", None) tracker_port = config.tracker_port or os.environ.get("TVM_TRACKER_PORT", None) tracker_key = config.tracker_key or os.environ.get("TVM_TRACKER_KEY", None) if isinstance(tracker_port, str): tracker_port = int(tracker_port) config = RPCConfig( tracker_host=tracker_host, tracker_port=tracker_port, tracker_key=tracker_key, session_priority=config.session_priority, session_timeout_sec=config.session_timeout_sec, ) config._sanity_check() # pylint: disable=protected-access return config def connect_tracker(self) -> rpc.TrackerSession: """Connect to the tracker Returns ------- tracker : TrackerSession The connected tracker session """ tracker: Optional[rpc.TrackerSession] = None def _connect(): nonlocal tracker tracker = rpc.connect_tracker(self.tracker_host, self.tracker_port) t = Thread(target=_connect) t.start() t.join(self.session_timeout_sec) if t.is_alive() or tracker is None: raise ValueError( "Unable to connect to the tracker using the following configuration:\n" f" tracker host: {self.tracker_host}\n" f" tracker port: {self.tracker_port}\n" f" timeout (sec): {self.session_timeout_sec}\n" "Please check the tracker status via the following command:\n" " python3 -m tvm.exec.query_rpc_tracker " f"--host {self.tracker_host} --port {self.tracker_port}" ) return tracker def connect_server(self) -> rpc.RPCSession: """Connect to the server Returns ------- session : RPCSession The connected rpc session """ tracker = self.connect_tracker() session: rpc.RPCSession = tracker.request( key=self.tracker_key, priority=self.session_priority, session_timeout=self.session_timeout_sec, ) return session def count_num_servers(self, allow_missing=True) -> int: """Count the number of servers available in the tracker Parameters ---------- allow_missing : bool Whether to allow no server to be found. Returns ------- num_servers : int The number of servers """ tracker = self.connect_tracker() tracker_summary = tracker.summary() result: int = 0 for item in tracker_summary["server_info"]: _, item_key = item["key"].split(":") if item_key == self.tracker_key: result += 1 if result == 0 and not allow_missing: raise ValueError( "Unable to find servers with the specific key using the following configuration:\n" f" tracker host: {self.tracker_host}\n" f" tracker port: {self.tracker_port}\n" f" tracker key: {self.tracker_key}\n" f" timeout (sec): {self.session_timeout_sec}\n" "Please check the tracker status via the following command:\n" " python3 -m tvm.exec.query_rpc_tracker " f"--host {self.tracker_host} --port {self.tracker_port}\n" f'and look for key: "{self.tracker_key}"' ) return result
7,373
35.686567
99
py
tvm
tvm-main/python/tvm/meta_schedule/runner/runner.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Runners""" from typing import Callable, List, Optional, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.runtime import Object from .. import _ffi_api from ..arg_info import ArgInfo @register_object("meta_schedule.RunnerInput") class RunnerInput(Object): """The runner's input Parameters ---------- artifact_path : str The path to the built artifact. device_type : str The device type. args_info : List[ArgInfo] The argument information. """ artifact_path: str device_type: str args_info: List[ArgInfo] def __init__( self, artifact_path: str, device_type: str, args_info: List[ArgInfo], ) -> None: """Constructor Parameters ---------- artifact_path : str The path to the built artifact. device_type : str The device type. args_info : List[ArgInfo] The argument information. """ self.__init_handle_by_constructor__( _ffi_api.RunnerInput, # type: ignore # pylint: disable=no-member artifact_path, device_type, args_info, ) @register_object("meta_schedule.RunnerResult") class RunnerResult(Object): """The runner's result Parameters ---------- run_secs : Optional[List[float]] The run time in seconds. error_msg : Optional[str] The error message, if any. """ run_secs: Optional[List[float]] error_msg: Optional[str] def __init__( self, run_secs: Optional[List[float]], error_msg: Optional[str], ) -> None: """Constructor Parameters ---------- run_secs : Optional[List[float]] The run time in seconds. error_msg : Optional[str] The error message, if any. """ self.__init_handle_by_constructor__( _ffi_api.RunnerResult, # type: ignore # pylint: disable=no-member run_secs, error_msg, ) @register_object("meta_schedule.RunnerFuture") class RunnerFuture(Object): """ A class to fetch asynchronous runner's output. This is NOT the user facing class for function overloading inheritance. Can be used for general return type of runner. See also: PyRunnerFuture """ def __init__(self, f_done: Callable, f_result: Callable = None) -> None: """Constructor""" self.__init_handle_by_constructor__( _ffi_api.RunnerFuture, # type: ignore # pylint: disable=no-member f_done, f_result, ) def done(self) -> bool: """Check whether the runner has finished.""" return _ffi_api.RunnerFutureDone(self) # type: ignore # pylint: disable=no-member def result(self) -> RunnerResult: """Fetch the runner's output if it is ready.""" return _ffi_api.RunnerFutureResult(self) # type: ignore # pylint: disable=no-member class PyRunnerFuture: """ A class to fetch asynchronous runner's output with customizable function on the python side. This is the user facing class for function overloading inheritance. Can NOT be used for general return type of runner. Note: @derived_object is required for proper usage of any inherited class. Example: @derived_object def LocalRunnerFuture(PyRunnerFuture): ... """ _tvm_metadata = { "cls": RunnerFuture, "methods": ["done", "result"], } def done(self) -> bool: """Check whether the runner has finished.""" raise NotImplementedError def result(self) -> RunnerResult: """Fetch the runner's output if it is ready.""" raise NotImplementedError @register_object("meta_schedule.Runner") class Runner(Object): """The abstract runner interface""" RunnerType = Union["Runner", Literal["local", "rpc"]] def run(self, runner_inputs: List[RunnerInput]) -> List[RunnerFuture]: """Run the built artifact and get runner futures. Parameters ---------- runner_inputs : List[RunnerInput] The inputs to the runner. Returns ------- runner_futures: List[RunnerFuture] The runner futures. """ return _ffi_api.RunnerRun(self, runner_inputs) # type: ignore # pylint: disable=no-member @staticmethod def create( # pylint: disable=keyword-arg-before-vararg kind: Literal["local", "rpc"] = "local", *args, **kwargs, ) -> "Runner": """Create a Runner.""" from . import LocalRunner, RPCRunner # pylint: disable=import-outside-toplevel if kind == "local": if "max_workers" in kwargs: kwargs.pop("max_workers") return LocalRunner(*args, **kwargs) # type: ignore elif kind == "rpc": return RPCRunner(*args, **kwargs) # type: ignore raise ValueError(f"Unknown Runner: {kind}") create = Runner.create # pylint: disable=invalid-name @register_object("meta_schedule.PyRunner") class _PyRunner(Runner): """ A TVM object runner to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyRunner """ def __init__(self, f_run: Callable = None) -> None: """Constructor""" self.__init_handle_by_constructor__( _ffi_api.RunnerPyRunner, # type: ignore # pylint: disable=no-member f_run, ) class PyRunner: """ An abstract runner with customized run method on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyRunner, "methods": ["run"], } def run(self, runner_inputs: List[RunnerInput]) -> List[RunnerFuture]: """Run the built artifact and get runner futures. Parameters ---------- runner_inputs : List[RunnerInput] The inputs to the runner. Returns ------- runner_futures: List[RunnerFuture] The runner futures. """ raise NotImplementedError
7,196
27.44664
98
py
tvm
tvm-main/python/tvm/meta_schedule/runner/rpc_runner.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """RPC Runner""" import concurrent.futures import os.path as osp from contextlib import contextmanager from typing import Callable, List, Optional, Union from tvm.contrib.popen_pool import PopenPoolExecutor from tvm.rpc import RPCSession from tvm.runtime import Device, Module from ..logging import get_logger from ..profiler import Profiler from ..utils import ( derived_object, get_global_func_on_rpc_session, get_global_func_with_default_on_worker, ) from .config import EvaluatorConfig, RPCConfig from .runner import PyRunner, PyRunnerFuture, RunnerFuture, RunnerInput, RunnerResult from .utils import ( T_ARG_INFO_JSON_OBJ_LIST, T_ARGUMENT_LIST, alloc_argument_common, run_evaluator_common, ) logger = get_logger(__name__) # pylint: disable=invalid-name T_CREATE_SESSION = Callable[ # pylint: disable=invalid-name [RPCConfig], # The RPC configuration RPCSession, # The RPC Session ] T_UPLOAD_MODULE = Callable[ # pylint: disable=invalid-name [ RPCSession, # The RPC Session str, # local path to the artifact str, # remote path to the artifact ], Module, # the Module opened on the remote ] T_ALLOC_ARGUMENT = Callable[ # pylint: disable=invalid-name [ RPCSession, # The RPC Session Device, # The device on the remote T_ARG_INFO_JSON_OBJ_LIST, # The metadata information of the arguments to be allocated int, # The number of repeated allocations to be done ], List[T_ARGUMENT_LIST], # A list of argument lists ] T_RUN_EVALUATOR = Callable[ # pylint: disable=invalid-name [ RPCSession, # The RPC Session Module, # The Module opened on the remote Device, # The device on the remote EvaluatorConfig, # The evaluator configuration List[T_ARGUMENT_LIST], # A list of argument lists ], List[float], # A list of running time ] T_CLEANUP = Callable[ # pylint: disable=invalid-name [ Optional[RPCSession], # The RPC Session to be cleaned up Optional[str], # remote path to the artifact ], None, ] @derived_object class RPCRunnerFuture(PyRunnerFuture): """RPC based runner future Parameters ---------- future: concurrent.futures.Future The concurrent function to check when the function is done and to return the result. timeout_sec: float The timeout in seconds. """ future: concurrent.futures.Future timeout_sec: float def __init__(self, future: concurrent.futures.Future, timeout_sec: float) -> None: """Constructor Parameters ---------- future: concurrent.futures.Future The concurrent function to check when the function is done and to return the result. timeout_sec: float The timeout in seconds. """ super().__init__() self.future = future self.timeout_sec = timeout_sec def done(self) -> bool: return self.future.done() def result(self) -> RunnerResult: try: run_secs: List[float] = self.future.result() except TimeoutError: return RunnerResult( None, error_msg=f"RPCRunner: Timeout, killed after {self.timeout_sec} seconds", ) except Exception as exception: # pylint: disable=broad-except return RunnerResult( None, error_msg="RPCRunner: An exception occurred\n" + str(exception), ) return RunnerResult(run_secs, None) @derived_object class RPCRunner(PyRunner): """RPC based runner Parameters ---------- rpc_config: RPCConfig The rpc configuration. evaluator_config: EvaluatorConfig The evaluator configuration. cooldown_sec: float The cooldown in seconds. TODO(@junrushao1994,@zxybazh): This is not used yet. alloc_repeat: int The number of times to repeat the allocation. f_create_session: Optional[str, Callable] The function name to create the session or the function itself. f_upload_module: Optional[str, Callable] The function name to upload the module or the function itself. f_alloc_argument: Optional[str, Callable] The function name to allocate the arguments or the function itself. f_run_evaluator: Optional[str, Callable] The function name to run the evaluator or the function itself. f_cleanup: Optional[str, Callable] The function name to cleanup the session or the function itself. pool: PopenPoolExecutor The popen pool executor. Attributes ---------- T_CREATE_SESSION : typing._GenericAlias The signature of the function `f_create_session`, which is: .. code-block:: python def default_create_session(rpc_config: RPCConfig) -> RPCSession: ... T_UPLOAD_MODULE : typing._GenericAlias The signature of the function `f_upload_module`, which is: .. code-block:: python def default_upload_module( session: RPCSession, local_path: str, remote_path: str, ) -> Module: ... T_ALLOC_ARGUMENT : typing._GenericAlias The signature of the function `f_alloc_argument`, which is: .. code-block:: python def default_alloc_argument( session: RPCSession, device: Device, args_info: T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[T_ARGUMENT_LIST]: ... T_RUN_EVALUATOR : typing._GenericAlias The signature of the function `f_run_evaluator`, which is: .. code-block:: python def default_run_evaluator( session: RPCSession, rt_mod: Module, device: Device, evaluator_config: EvaluatorConfig, repeated_args: List[T_ARGUMENT_LIST], ) -> List[float]: ... T_CLEANUP : typing._GenericAlias The signature of the function `f_cleanup`, which is: .. code-block:: python def default_cleanup( session: Optional[RPCSession], remote_path: Optional[str], ) -> None: ... """ rpc_config: RPCConfig evaluator_config: EvaluatorConfig cooldown_sec: float alloc_repeat: int f_create_session: Union[T_CREATE_SESSION, str, None] f_upload_module: Union[T_UPLOAD_MODULE, str, None] f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] f_cleanup: Union[T_CLEANUP, str, None] pool: PopenPoolExecutor def __init__( self, rpc_config: Optional[RPCConfig] = None, evaluator_config: Optional[EvaluatorConfig] = None, cooldown_sec: float = 0.0, alloc_repeat: int = 1, f_create_session: Union[T_CREATE_SESSION, str, None] = None, f_upload_module: Union[T_UPLOAD_MODULE, str, None] = None, f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] = None, f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] = None, f_cleanup: Union[T_CLEANUP, str, None] = None, max_workers: Optional[int] = None, initializer: Optional[Callable[[], None]] = None, ) -> None: """Constructor Parameters ---------- rpc_config: RPCConfig The rpc configuration. evaluator_config: EvaluatorConfig The evaluator configuration. cooldown_sec: float The cooldown in seconds. alloc_repeat: int The number of times to random fill the allocation. f_create_session: Union[T_CREATE_SESSION, str, None] The function name to create the session or the function itself. f_upload_module: Union[T_UPLOAD_MODULE, str, None] The function name to upload the module or the function itself. f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] The function name to allocate the arguments or the function itself. f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] The function name to run the evaluator or the function itself. f_cleanup: Union[T_CLEANUP, str, None] The function name to cleanup the session or the function itself. max_workers: Optional[int] = None The maximum number of connections. Defaults to 1. initializer: Optional[Callable[[], None]] The initializer function. """ super().__init__() self.rpc_config = RPCConfig._normalized(rpc_config) self.evaluator_config = EvaluatorConfig._normalized(evaluator_config) self.cooldown_sec = cooldown_sec self.alloc_repeat = alloc_repeat self.f_create_session = f_create_session self.f_upload_module = f_upload_module self.f_alloc_argument = f_alloc_argument self.f_run_evaluator = f_run_evaluator self.f_cleanup = f_cleanup if max_workers is None: max_workers = 1 logger.info("RPCRunner: max_workers = %d", max_workers) self.pool = PopenPoolExecutor( max_workers=max_workers, initializer=initializer, ) self._sanity_check() def run(self, runner_inputs: List[RunnerInput]) -> List[RunnerFuture]: results: List[RunnerFuture] = [] for runner_input in runner_inputs: future = RPCRunnerFuture( future=self.pool.submit( _worker_func, self.f_create_session, self.f_upload_module, self.f_alloc_argument, self.f_run_evaluator, self.f_cleanup, self.rpc_config, self.evaluator_config, self.alloc_repeat, str(runner_input.artifact_path), str(runner_input.device_type), tuple(arg_info.as_json() for arg_info in runner_input.args_info), ), timeout_sec=self.rpc_config.session_timeout_sec, ) results.append(future) # type: ignore return results def _sanity_check(self) -> None: def _check( f_create_session, f_upload_module, f_alloc_argument, f_run_evaluator, f_cleanup, ) -> None: get_global_func_with_default_on_worker(name=f_create_session, default=None) get_global_func_with_default_on_worker(name=f_upload_module, default=None) get_global_func_with_default_on_worker(name=f_alloc_argument, default=None) get_global_func_with_default_on_worker(name=f_run_evaluator, default=None) get_global_func_with_default_on_worker(name=f_cleanup, default=None) value = self.pool.submit( _check, self.f_create_session, self.f_upload_module, self.f_alloc_argument, self.f_run_evaluator, self.f_cleanup, ) value.result() def _worker_func( _f_create_session: Union[T_CREATE_SESSION, str, None], _f_upload_module: Union[T_UPLOAD_MODULE, str, None], _f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None], _f_run_evaluator: Union[T_RUN_EVALUATOR, str, None], _f_cleanup: Union[T_CLEANUP, str, None], rpc_config: RPCConfig, evaluator_config: EvaluatorConfig, alloc_repeat: int, artifact_path: str, device_type: str, args_info: T_ARG_INFO_JSON_OBJ_LIST, ) -> List[float]: # Step 0. Get the registered functions f_create_session: T_CREATE_SESSION = get_global_func_with_default_on_worker( _f_create_session, default_create_session ) f_upload_module: T_UPLOAD_MODULE = get_global_func_with_default_on_worker( _f_upload_module, default_upload_module ) f_alloc_argument: T_ALLOC_ARGUMENT = get_global_func_with_default_on_worker( _f_alloc_argument, default_alloc_argument ) f_run_evaluator: T_RUN_EVALUATOR = get_global_func_with_default_on_worker( _f_run_evaluator, default_run_evaluator ) f_cleanup: T_CLEANUP = get_global_func_with_default_on_worker(_f_cleanup, default_cleanup) # Managed resources session: Optional[RPCSession] = None remote_path: Optional[str] = None @contextmanager def resource_handler(): try: yield finally: # Final step. Always clean up with Profiler.timeit("RPCRunner/cleanup"): f_cleanup(session, remote_path) with resource_handler(): # Step 1. Create session with Profiler.timeit("RPCRunner/create_session"): session = f_create_session(rpc_config) device = session.device(dev_type=device_type, dev_id=0) # Step 2. Upload the module with Profiler.timeit("RPCRunner/upload_module"): _, remote_path = osp.split(artifact_path) local_path: str = artifact_path rt_mod: Module = f_upload_module(session, local_path, remote_path) # Step 3: Allocate input arguments with Profiler.timeit("RPCRunner/alloc_argument"): repeated_args: List[T_ARGUMENT_LIST] = f_alloc_argument( session, device, args_info, alloc_repeat, ) # Step 4: Run time_evaluator with Profiler.timeit("LocalRunner/run_evaluator"): costs: List[float] = f_run_evaluator( session, rt_mod, device, evaluator_config, repeated_args, ) return costs def default_create_session(rpc_config: RPCConfig) -> RPCSession: """Default function to create the session Parameters ---------- rpc_config : RPCConfig The configuration of the RPC session Returns ------- session : RPCSession The created rpc session """ return rpc_config.connect_server() def default_upload_module( session: RPCSession, local_path: str, remote_path: str, ) -> Module: """Default function to upload the module Parameters ---------- session: RPCSession The session to upload the module local_path: str The local path of the module remote_path: str The remote path to place the module Returns ------- rt_mod : Module The runtime module """ session.upload(local_path, remote_path) rt_mod: Module = session.load_module(remote_path) return rt_mod def default_alloc_argument( session: RPCSession, device: Device, args_info: T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[T_ARGUMENT_LIST]: """Default function to allocate the arguments Parameters ---------- session: RPCSession The session to allocate the arguments device: Device The device to allocate the arguments args_info: T_ARG_INFO_JSON_OBJ_LIST The arguments info alloc_repeat: int The number of times to repeat the allocation Returns ------- repeated_args: List[Args] The allocation args """ f_random_fill = get_global_func_on_rpc_session( session, "tvm.contrib.random.random_fill_for_measure", "Please make sure 'USE_RANDOM' is turned ON in the config.cmake on the RPC server.", ) return alloc_argument_common(f_random_fill, device, args_info, alloc_repeat) def default_run_evaluator( session: RPCSession, # pylint: disable=unused-argument rt_mod: Module, device: Device, evaluator_config: EvaluatorConfig, repeated_args: List[T_ARGUMENT_LIST], ) -> List[float]: """Default function to run the evaluator Parameters ---------- session: RPCSession The session to run the evaluator rt_mod: Module The runtime module device: Device The device to run the evaluator evaluator_config: EvaluatorConfig The evaluator config repeated_args: List[T_ARGUMENT_LIST] The repeated arguments Returns ------- costs: List[float] The evaluator results """ return run_evaluator_common(rt_mod, device, evaluator_config, repeated_args) def default_cleanup( session: Optional[RPCSession], remote_path: Optional[str], ) -> None: """Default function to clean up the session Parameters ---------- session: RPCSession The session to clean up remote_path: str The remote path to clean up """ if session is not None and remote_path is not None: session.remove(remote_path) session.remove(remote_path + ".so") session.remove("")
17,682
32.052336
96
py
tvm
tvm-main/python/tvm/meta_schedule/runner/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.runner package. Meta Schedule runners that runs an artifact either locally or through the RPC interface """ from .config import EvaluatorConfig, RPCConfig from .local_runner import LocalRunner, LocalRunnerFuture from .rpc_runner import RPCRunner from .runner import ( PyRunner, PyRunnerFuture, Runner, RunnerFuture, RunnerInput, RunnerResult, create, )
1,192
35.151515
87
py
tvm
tvm-main/python/tvm/meta_schedule/runner/local_runner.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Local Runner""" from contextlib import contextmanager from typing import Callable, List, Optional, Union import subprocess import tvm from ...contrib.popen_pool import PopenPoolExecutor from ...runtime import Device, Module from ..logging import get_logger from ..profiler import Profiler from ..utils import derived_object, get_global_func_with_default_on_worker from .config import EvaluatorConfig from .runner import PyRunner, PyRunnerFuture, RunnerFuture, RunnerInput, RunnerResult from .utils import ( T_ARG_INFO_JSON_OBJ_LIST, T_ARGUMENT_LIST, alloc_argument_common, run_evaluator_common, ) logger = get_logger(__name__) # pylint: disable=invalid-name T_ALLOC_ARGUMENT = Callable[ # pylint: disable=invalid-name [ Device, # The device on the remote T_ARG_INFO_JSON_OBJ_LIST, # The metadata information of the arguments to be allocated int, # The number of repeated allocations to be done ], List[T_ARGUMENT_LIST], # A list of argument lists ] T_RUN_EVALUATOR = Callable[ # pylint: disable=invalid-name [ Module, # The Module opened on the remote Device, # The device on the remote EvaluatorConfig, # The evaluator configuration List[T_ARGUMENT_LIST], # A list of argument lists ], List[float], # A list of running time ] T_CLEANUP = Callable[ # pylint: disable=invalid-name [], None, ] @derived_object class LocalRunnerFuture(PyRunnerFuture): """Local based runner future Parameters ---------- res: Optional[List[float]] The optional result as a list of float. error_message: Optional[str] The optional error message. Note ---- Only one of the parameters should be None upon the creation of LocalRunnerFuture object """ res: Optional[List[float]] error_message: Optional[str] def __init__( self, res: Optional[List[float]] = None, error_message: Optional[str] = None ) -> None: """Constructor Parameters ---------- res: Optional[List[float]] The result of this LocalRunnerFuture error_message: Optional[str] The stringfied error message of any exception during execution """ super().__init__() self.res = res self.error_message = error_message # sanity check upon the creation of LocalRunnerFuture object if (res is None and error_message is None) or ( res is not None and error_message is not None ): raise AttributeError( "Only one of the two parameters should be None upon the creation" "of LocalRunnerFuture object." ) def done(self) -> bool: return True def result(self) -> RunnerResult: return RunnerResult(self.res, self.error_message) def _worker_func( _f_alloc_argument: Optional[str], _f_run_evaluator: Optional[str], _f_cleanup: Optional[str], evaluator_config: EvaluatorConfig, alloc_repeat: int, artifact_path: str, device_type: str, args_info: T_ARG_INFO_JSON_OBJ_LIST, ) -> List[float]: f_alloc_argument: T_ALLOC_ARGUMENT = get_global_func_with_default_on_worker( _f_alloc_argument, default_alloc_argument ) f_run_evaluator: T_RUN_EVALUATOR = get_global_func_with_default_on_worker( _f_run_evaluator, default_run_evaluator ) f_cleanup: T_CLEANUP = get_global_func_with_default_on_worker(_f_cleanup, default_cleanup) @contextmanager def resource_handler(): try: yield finally: # Final step. Always clean up with Profiler.timeit("LocalRunner/cleanup"): f_cleanup() with resource_handler(): # Step 1: create the local runtime module with Profiler.timeit("LocalRunner/load_module"): rt_mod = tvm.runtime.load_module(artifact_path) # Step 2: Allocate input arguments with Profiler.timeit("LocalRunner/alloc_argument"): device = tvm.runtime.device(dev_type=device_type, dev_id=0) repeated_args: List[T_ARGUMENT_LIST] = f_alloc_argument( device, args_info, alloc_repeat, ) # Step 3: Run time_evaluator with Profiler.timeit("LocalRunner/run_evaluator"): costs: List[float] = f_run_evaluator( rt_mod, device, evaluator_config, repeated_args, ) return costs @derived_object class LocalRunner(PyRunner): """Local runner Parameters ---------- evaluator_config: EvaluatorConfig The evaluator configuration. cooldown_sec: float The cooldown in seconds. alloc_repeat: int The number of times to repeat the allocation. f_alloc_argument: Optional[str, Callable] The function name to allocate the arguments or the function itself. f_run_evaluator: Optional[str, Callable] The function name to run the evaluator or the function itself. f_cleanup: Optional[str, Callable] The function name to cleanup the session or the function itself. pool: PopenPoolExecutor The popen pool executor. Attributes ---------- T_ALLOC_ARGUMENT : typing._GenericAlias The signature of the function `f_alloc_argument`, which is: .. code-block:: python def default_alloc_argument( device: Device, args_info: T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[T_ARGUMENT_LIST]: ... T_RUN_EVALUATOR : typing._GenericAlias The signature of the function `f_run_evaluator`, which is: .. code-block:: python def default_run_evaluator( rt_mod: Module, device: Device, evaluator_config: EvaluatorConfig, repeated_args: List[T_ARGUMENT_LIST], ) -> List[float]: ... T_CLEANUP : typing._GenericAlias The signature of the function `f_cleanup`, which is: .. code-block:: python def default_cleanup() -> None: ... """ timeout_sec: float evaluator_config: EvaluatorConfig cooldown_sec: float alloc_repeat: int f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] f_cleanup: Union[T_CLEANUP, str, None] pool: PopenPoolExecutor def __init__( self, timeout_sec: float = 30, evaluator_config: Optional[EvaluatorConfig] = None, cooldown_sec: float = 0.0, alloc_repeat: int = 1, f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] = None, f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] = None, f_cleanup: Union[T_CLEANUP, str, None] = None, initializer: Optional[Callable[[], None]] = None, ) -> None: """Constructor Parameters ---------- timeout_sec: float The timeout setting. evaluator_config: EvaluatorConfig The evaluator configuration. cooldown_sec: float The cooldown in seconds. alloc_repeat: int The number of times to random fill the allocation. f_alloc_argument: Union[T_ALLOC_ARGUMENT, str, None] The function name to allocate the arguments or the function itself. f_run_evaluator: Union[T_RUN_EVALUATOR, str, None] The function name to run the evaluator or the function itself. f_cleanup: Union[T_CLEANUP, str, None] The function name to cleanup the session or the function itself. initializer: Optional[Callable[[], None]] The initializer function. """ super().__init__() self.timeout_sec = timeout_sec self.evaluator_config = EvaluatorConfig._normalized(evaluator_config) self.cooldown_sec = cooldown_sec self.alloc_repeat = alloc_repeat self.f_alloc_argument = f_alloc_argument self.f_run_evaluator = f_run_evaluator self.f_cleanup = f_cleanup logger.info("LocalRunner: max_workers = 1") self.pool = PopenPoolExecutor( max_workers=1, # one local worker timeout=timeout_sec, initializer=initializer, stderr=subprocess.DEVNULL, # suppress the stderr output ) self._sanity_check() def run(self, runner_inputs: List[RunnerInput]) -> List[RunnerFuture]: results: List[RunnerFuture] = [] for runner_input in runner_inputs: future = self.pool.submit( _worker_func, self.f_alloc_argument, self.f_run_evaluator, self.f_cleanup, self.evaluator_config, self.alloc_repeat, str(runner_input.artifact_path), str(runner_input.device_type), tuple(arg_info.as_json() for arg_info in runner_input.args_info), ) try: result: List[float] = future.result() error_message: str = None except TimeoutError: result = None error_message = f"LocalRunner: Timeout, killed after {self.timeout_sec} seconds\n" except Exception as exception: # pylint: disable=broad-except result = None error_message = "LocalRunner: An exception occurred\n" + str(exception) local_future = LocalRunnerFuture(res=result, error_message=error_message) results.append(local_future) # type: ignore return results def _sanity_check(self) -> None: def _check( f_alloc_argument, f_run_evaluator, f_cleanup, ) -> None: get_global_func_with_default_on_worker(name=f_alloc_argument, default=None) get_global_func_with_default_on_worker(name=f_run_evaluator, default=None) get_global_func_with_default_on_worker(name=f_cleanup, default=None) value = self.pool.submit( _check, self.f_alloc_argument, self.f_run_evaluator, self.f_cleanup, ) value.result() def default_alloc_argument( device: Device, args_info: T_ARG_INFO_JSON_OBJ_LIST, alloc_repeat: int, ) -> List[T_ARGUMENT_LIST]: """Default function to allocate the arguments Parameters ---------- device: Device The device to allocate the arguments args_info: T_ARG_INFO_JSON_OBJ_LIST The arguments info alloc_repeat: int The number of times to repeat the allocation Returns ------- repeated_args: List[T_ARGUMENT_LIST] The allocation args """ f_random_fill = get_global_func_with_default_on_worker( name="tvm.contrib.random.random_fill_for_measure", default=None ) return alloc_argument_common(f_random_fill, device, args_info, alloc_repeat) def default_run_evaluator( rt_mod: Module, device: Device, evaluator_config: EvaluatorConfig, repeated_args: List[T_ARGUMENT_LIST], ) -> List[float]: """Default function to run the evaluator Parameters ---------- rt_mod: Module The runtime module device: Device The device to run the evaluator evaluator_config: EvaluatorConfig The evaluator config repeated_args: List[T_ARGUMENT_LIST] The repeated arguments Returns ------- costs: List[float] The evaluator results """ return run_evaluator_common(rt_mod, device, evaluator_config, repeated_args) def default_cleanup() -> None: """Default function to clean up the session""" pass # pylint: disable=unnecessary-pass
12,681
31.685567
98
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/mlp_model.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # type: ignore[import] """ Segment Sum MLP cost model """ import glob import math import os import random import tempfile from collections import OrderedDict from itertools import chain as itertools_chain from typing import Dict, List, NamedTuple, Optional, Tuple import numpy as np # type: ignore import torch # type: ignore import tvm from ...contrib.tar import tar, untar from ...runtime import NDArray from ...target import Target from ..cost_model import PyCostModel from ..database import JSONDatabase from ..feature_extractor import FeatureExtractor, PerStoreFeature from ..logging import get_logger from ..runner import RunnerResult from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext from ..utils import derived_object, shash2hex logger = get_logger("mlp_model") # pylint: disable=invalid-name # pylint: disable=no-member,import-outside-toplevel class SegmentSumMLPConfig(NamedTuple): """SegmentSum MLP model configuration Parameters ---------- input_dim : int The input dim for the model. hidden_dim : int The hidden dim for the model. output_dim : int The output dim for the model. use_norm : bool Whether to normalize the segment sum or not. use_sigmoid : bool Whether to use sigmoid on the final output or not. """ input_dim: int = 172 hidden_dim: int = 256 output_dim: int = 1 use_norm: bool = False use_sigmoid: bool = False def to_dict(self): # pylint: disable=missing-function-docstring return { "input_dim": self.input_dim, "hidden_dim": self.hidden_dim, "output_dim": self.output_dim, "use_norm": self.use_norm, "use_sigmoid": self.use_sigmoid, } class TrainerConfig(NamedTuple): """Trainer configuration Parameters ---------- batch_size : int The batch size. learning rate : float The learning rate. weight decay : float The weight decay. num_epoch_full : int The number of epochs used in full training. num_epoch_incremental : int The number of epochs used in incremental training. grad_clip_norm: float The norm of gradient clipping. train_verbose: int The verbose frequency for training in batches. test_interval: int The testing interval in epochs. test_split: float The fraction of data for testing. frozen: bool Determine whether to re-train the model or not. """ batch_size: int = 128 learning_rate: float = 7e-4 weight_decay: float = 1e-6 num_epoch_full: int = 50 num_epoch_incremental: int = 5 grad_clip_norm: float = 0.5 train_verbose: int = 1000 test_interval: int = 1 test_split: float = 0.2 frozen: bool = False def to_dict(self): # pylint: disable=missing-function-docstring return { "batch_size": self.batch_size, "learning_rate": self.learning_rate, "weight_decay": self.weight_decay, "num_epoch_full": self.num_epoch_full, "num_epoch_incremental": self.num_epoch_incremental, "grad_clip_norm": self.grad_clip_norm, "train_verbose": self.train_verbose, "test_interval": self.test_interval, "test_split": self.test_split, "frozen": self.frozen, } # pylint: disable=too-few-public-methods class FeatureGroup: """Feature group Parameters ---------- group_hash : str The hash of the group features : List[np.ndarray] The features costs : List[float] The costs min_cost : float The minimum cost """ group_hash: str features: List[np.ndarray] costs: np.ndarray min_cost: float def __init__( self, group_hash: str, features: List[np.ndarray], costs: np.ndarray, ) -> None: self.group_hash = group_hash self.features = features self.costs = costs self.min_cost = np.min(costs) def append( # pylint: disable=missing-function-docstring self, features: List[np.ndarray], costs: np.ndarray, ) -> None: self.features.extend(features) self.costs = np.append(self.costs, costs) self.min_cost = np.min(self.costs) # pylint: disable=too-many-instance-attributes class SegmentDataLoader: """Dataloader for Segment Sum MLP model. Parameters ---------- features : List[np.ndarray] The features results : np.ndarray The measured results, can be None. batch_size : int The batch size shuffle : bool Whether to shuffle the dataset or not """ def __init__( self, features, results=None, batch_size=128, shuffle=True, ): self.batch_size = batch_size self.shuffle = shuffle self.data_size = len(features) # flatten features and store the starting indices self.segment_sizes = torch.tensor([len(feature) for feature in features], dtype=torch.int32) self.feature_offsets = ( torch.cumsum(self.segment_sizes, 0, dtype=torch.int32) - self.segment_sizes ) features = torch.cat([torch.tensor(feature) for feature in features]) norm, _ = features.max(dim=0) norm[norm == 0] = 1 self.features = features / norm self.results = torch.tensor(results) if results is not None else None self.iter_order = self.pointer = None def __len__(self): return self.data_size def __iter__(self): if self.shuffle: self.iter_order = torch.randperm(self.data_size) else: self.iter_order = torch.arange(self.data_size) self.pointer = 0 return self def __next__(self): if self.pointer >= self.data_size: raise StopIteration batch_indices = self.iter_order[self.pointer : self.pointer + self.batch_size] self.pointer += self.batch_size return self._fetch_indices(batch_indices) def _fetch_indices(self, indices): segment_sizes, feature_offsets = self.segment_sizes[indices], self.feature_offsets[indices] feature_indices = torch.empty(segment_sizes.sum(), dtype=torch.int32) idx = 0 for offset, seg_size in zip(feature_offsets, segment_sizes): feature_indices[idx : idx + seg_size] = torch.arange(offset, offset + seg_size) idx += seg_size features = self.features[feature_indices.long()] results = None if self.results is not None: results = self.results[indices.long()] return segment_sizes, features, results def lambda_rank_loss( # pylint: disable=too-many-locals preds: "torch.Tensor", labels: "torch.Tensor", k: int = None, eps: float = 1e-10, sigma: float = 1.0, ) -> "torch.Tensor": """ LambdaLoss: Metric-Driven Loss for Learning-to-Rank Parameters ---------- preds : Tensor The predicted runtime for each candidate. labels : Tensor The measured runtime for each candidate. k : int Loss for top k. Default is None, which means computing all scores. eps : float The minimum value to the denominator and argument of log if they reach 0. sigma : float The scaling factor to the input of the sigmoid function. Returns ------- loss : Tensor The lambda rank loss. """ device = preds.device y_pred, y_true = preds[None, :], labels[None, :] y_pred_sorted, indices_pred = y_pred.sort(descending=True, dim=-1) y_true_sorted, _ = y_true.sort(descending=True, dim=-1) true_sorted_by_preds = torch.gather(y_true, dim=1, index=indices_pred) true_diffs = true_sorted_by_preds[:, :, None] - true_sorted_by_preds[:, None, :] padded_pairs_mask = torch.isfinite(true_diffs) & (true_diffs > 0) ndcg_at_k_mask = torch.zeros( (y_pred.shape[1], y_pred.shape[1]), dtype=torch.bool, device=device ) ndcg_at_k_mask[:k, :k] = 1 true_sorted_by_preds.clamp_(min=0.0) y_true_sorted.clamp_(min=0.0) pos_idxs = torch.arange(1, y_pred.shape[1] + 1).to(device) D = torch.log2(1.0 + pos_idxs.float())[None, :] # pylint: disable=invalid-name maxDCGs = torch.sum( # pylint: disable=invalid-name ((torch.pow(2, y_true_sorted) - 1) / D)[:, :k], dim=-1 ).clamp(min=eps) G = (torch.pow(2, true_sorted_by_preds) - 1) / maxDCGs[:, None] # pylint: disable=invalid-name weights = torch.abs( torch.pow(D[:, :, None], -1.0) - torch.pow(D[:, None, :], -1.0) ) * torch.abs(G[:, :, None] - G[:, None, :]) scores_diffs = (y_pred_sorted[:, :, None] - y_pred_sorted[:, None, :]).clamp(min=-1e8, max=1e8) scores_diffs[torch.isnan(scores_diffs)] = 0.0 weighted_probs = (torch.sigmoid(sigma * scores_diffs).clamp(min=eps) ** weights).clamp(min=eps) losses = torch.log2(weighted_probs) masked_losses = losses[padded_pairs_mask & ndcg_at_k_mask] loss = -torch.sum(masked_losses) return loss def topk_score( pred_results: "torch.Tensor", gt_results: "torch.Tensor", k: int, ) -> float: """ Evaluate the top-k score Parameters ---------- pred_results: Tensor The raw prediction gt_results: Tensor The measured results k : int The k in top k score Returns ------- score : float The top-k score """ k = min(k, len(pred_results)) topk_indices = torch.topk(pred_results, k, largest=False).indices score = gt_results.min() / gt_results[topk_indices].min() return score.item() class SegmentSumMLP(torch.nn.Module): """Segment Sum MLP model. Parameters ---------- input_dim : int The input dim for the model. hidden_dim : int The hidden dim for the model. output_dim : int The output dim for the model. use_norm : bool Whether to normalize the segment sum or not. use_sigmoid : bool Whether to use sigmoid on the final output or not. """ input_dim: int hidden_dim: int output_dim: int use_norm: bool use_sigmoid: bool def __init__( # pylint: disable=too-many-arguments self, input_dim: int = 172, hidden_dim: int = 256, output_dim: int = 1, use_norm: bool = False, use_sigmoid: bool = False, ): from torch import nn # type: ignore super().__init__() self.encoder = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.ReLU(), nn.Linear(hidden_dim, hidden_dim), nn.ReLU(), ) self.norm = nn.BatchNorm1d(hidden_dim) if use_norm else nn.Identity() self.layer0 = nn.Sequential( nn.Linear(hidden_dim, hidden_dim), nn.ReLU(), ) self.layer1 = nn.Sequential( nn.Linear(hidden_dim, hidden_dim), nn.ReLU(), ) self.decoder = nn.Linear(hidden_dim, output_dim) self.sigmoid = nn.Sigmoid() if use_sigmoid else nn.Identity() def forward( # pylint: disable=missing-function-docstring self, segment_sizes: "torch.Tensor", features: "torch.Tensor", ) -> "torch.Tensor": n_seg = len(segment_sizes) encoded_features = self.encoder(features) segment_indices = torch.repeat_interleave( torch.arange(n_seg, device=features.device), segment_sizes.long(), ) n_dim = encoded_features.shape[1] segment_sum = torch.scatter_add( input=torch.zeros((n_seg, n_dim), dtype=encoded_features.dtype, device=features.device), dim=0, index=segment_indices.view(-1, 1).expand(-1, n_dim), src=encoded_features, ) out = self.norm(segment_sum) out = self.layer0(out) + out out = self.layer1(out) + out out = self.decoder(out).squeeze() out = self.sigmoid(out) return out def extract_features( context: TuneContext, candidates: List[MeasureCandidate], results: Optional[List[RunnerResult]] = None, extractor: Optional[FeatureExtractor] = None, ): """Extract feature vectors and compute mean costs. Parameters ---------- context: TuneContext The tuning context. candidates: List[MeasureCandidate] The measure candidates. results: Optional[List[RunnerResult]] The measured results, can be None if used in prediction. extractor: Optional[FeatureExtractor] The feature extractor. Returns ------- new_features: List[np.ndarray] The extracted features. new_mean_costs: np.ndarray The mean costs. """ extractor = extractor or PerStoreFeature(extract_workload=True) def _feature(feature: NDArray) -> np.ndarray: return feature.numpy().astype("float32") def _mean_cost(res: RunnerResult) -> float: if not res.run_secs: return 1e10 return float(np.median([float(s) for s in res.run_secs])) new_features = [_feature(x) for x in extractor.extract_from(context, candidates)] new_mean_costs = ( np.array([_mean_cost(x) for x in results]).astype("float32") if results is not None else None ) return new_features, new_mean_costs class State: """State of the trainer Parameters ---------- model: SegmentSumMLP The cost model. data: Dict[str, FeatureGroup] The data groups. data_size: int The size of all data. untrained_size: int The size of the untrained data. """ model: SegmentSumMLP data: Dict[str, FeatureGroup] data_size: int untrained_size: int def __init__( self, model_config: Optional[SegmentSumMLPConfig] = None, extractor: Optional[FeatureExtractor] = None, ): model_config = model_config or SegmentSumMLPConfig() extractor = extractor or PerStoreFeature(extract_workload=True) self.model = SegmentSumMLP(**model_config.to_dict()) self.data = OrderedDict() self.data_size = 0 self.untrained_size = 0 self.extractor = extractor def load( # pylint: disable=too-many-locals self, path: str, target: str = "nvidia/nvidia-v100", ) -> None: """Load the cached model, cached features, or raw data. Parameters ---------- path: str The path to the tar file containing cached model, cached features, or raw data. target: str The target for the tuning context. """ with tempfile.TemporaryDirectory() as tmp_dir: model_path = os.path.join(tmp_dir, "model.pth") cache_path = os.path.join(tmp_dir, "cached_data.npy") raw_path = os.path.join(tmp_dir, "raw_data") untar(path, tmp_dir) if os.path.exists(model_path): self.model.load_state_dict(torch.load(model_path)) if os.path.exists(cache_path): for group_hash, features, costs in np.load(cache_path, allow_pickle=True): self.data[group_hash] = FeatureGroup( group_hash=group_hash, features=list(features), costs=costs, ) self.data_size += len(costs) self.untrained_size += len(costs) elif os.path.exists(raw_path): from tqdm import tqdm # type: ignore model_dirs = glob.glob(os.path.join(raw_path, "*")) workload_paths = [] for model_dir in model_dirs: json_files = glob.glob(os.path.join(model_dir, "*.json")) for json_file in json_files: if json_file.endswith("_workload.json"): workload_paths.append(json_file) for workload_path in tqdm(workload_paths): try: database = JSONDatabase( path_workload=workload_path, path_tuning_record=workload_path.replace( "_workload.json", "_candidates.json" ), ) except tvm._ffi.base.TVMError: # pylint: disable=protected-access continue candidates, results = [], [] tuning_records = database.get_all_tuning_records() if len(tuning_records) == 0: continue for record in tuning_records: candidates.append(record.as_measure_candidate()) results.append(RunnerResult(run_secs=record.run_secs, error_msg=None)) assert len(candidates) == len(results) context = TuneContext(mod=tuning_records[0].workload.mod, target=Target(target)) features, mean_costs = extract_features( context, candidates, results, self.extractor ) self.add_to_group(features, mean_costs, shash2hex(context.mod)) def save(self, path: str) -> None: """Cache the model and data. Parameters ---------- path: str The path to the cached tar file. """ with tempfile.TemporaryDirectory() as tmp_dir: model_path = os.path.join(tmp_dir, "model.pth") cache_path = os.path.join(tmp_dir, "cached_data.npy") torch.save(self.model.state_dict(), model_path) data = [ ( g.group_hash, g.features, g.costs, ) for g in self.data.values() ] np.save( file=cache_path, arr=np.array(data, dtype=object), ) tar(path, [x for x in [model_path, cache_path] if x is not None]) logger.info("Saved MLPModel to %s", path) def add_to_group( self, features: List[np.ndarray], costs: np.ndarray, group_hash: str, ): """Add features and costs to the data groups with key group_hash. Parameters ---------- features: List[np.ndarray] The feature vectors. costs: np.ndarray The measured results. group_hash: str The structural hash of the candidates. """ group = self.data.get(group_hash, None) if group is None: group = FeatureGroup( group_hash=group_hash, features=features, costs=costs, ) else: group.append(features, costs) self.data[group_hash] = group self.data_size += len(features) self.untrained_size += len(features) class SegmentSumMLPTrainer: """The trainer for Segment Sum MLP model. Parameters ---------- state: State The state of the trainer. batch_size : int The batch size. learning rate : float The learning rate. weight decay : float The weight decay. num_epoch_full : int The number of epochs used in full training. num_epoch_incremental : int The number of epochs used in incremental training. grad_clip_norm: float The norm of gradient clipping. train_verbose: int The verbose frequency for training in batches. test_interval: int The testing interval in epochs. test_split: float The fraction of data for testing. frozen: bool Determine whether to re-train the model or not. optimizer: "torch.optim.adam.Adam" The optimizer. scheduler: "torch.optim.lr_scheduler.StepLR" The scheduler. """ state: State batch_size: int = 128 learning_rate: float = 7e-4 weight_decay: float = 1e-6 num_epoch_full: int = 50 num_epoch_incremental: int = 5 grad_clip_norm: float = 0.5 train_verbose: int = 1000 test_interval: int = 1 test_split: float = 0.2 frozen: bool = False optimizer: "torch.optim.adam.Adam" # type: ignore scheduler: "torch.optim.lr_scheduler.StepLR" # type: ignore def __init__( self, train_config: Optional[TrainerConfig] = None, state: Optional[State] = None, ): train_config = train_config or TrainerConfig() state = state or State() config = train_config.to_dict() for attr in config: setattr(self, attr, config[attr]) self.state = state self.device = "cuda" if torch.cuda.device_count() else "cpu" self.optimizer, self.scheduler = None, None def train_step( self, data: Tuple["torch.Tensor", "torch.Tensor", "torch.Tensor"], batch: int = 0, train_loss: Optional[float] = None, ) -> float: """Helper function for training on a single batch. Parameters ---------- data: Tuple["torch.Tensor", "torch.Tensor", "torch.Tensor"] A batch of data, should be a tuple of (segment_sizes, features, gt_results). batch: int = 0 The current batch number. train_loss: Optional[float] = None The previous averaged training loss, None if it is the first batch. Returns ------- train_loss: float The averaged training loss after the current batch. """ segment_sizes, features, gt_results = ( data[0].to(self.device), data[1].to(self.device), data[2].to(self.device), ) self.optimizer.zero_grad() pred_results = self.state.model(segment_sizes, features) loss = lambda_rank_loss(pred_results, gt_results) loss.backward() torch.nn.utils.clip_grad_norm_(self.state.model.parameters(), self.grad_clip_norm) self.optimizer.step() loss = loss.detach().cpu() train_loss = ( train_loss * 0.95 + loss.item() * 0.05 if train_loss is not None else loss.item() ) segment_sizes, features, gt_results, pred_results = ( segment_sizes.detach().cpu(), features.detach().cpu(), gt_results.detach().cpu(), pred_results.detach().cpu(), ) if batch % self.train_verbose == 0: logger.info("Batch: %d, train loss: %6f", batch, train_loss) return train_loss def predict_step( self, data: Tuple["torch.Tensor", "torch.Tensor", "torch.Tensor"], ): """Helper function for predicting (validating) on a single batch. Parameters ---------- data: Tuple["torch.Tensor", "torch.Tensor", "torch.Tensor"] A batch of data, should be a tuple of (segment_sizes, features, gt_results). gt_results can be None if it is used for predicting. Returns ------- pred_results: np.ndarray The predicted results for the current batch. test_loss_batch: float If used for validation, return the test loss for the current batch. test_scores_batch: List[float] If used for validation, return the topk scores for the current batch. """ test_loss_batch, test_scores_batch = None, [] segment_sizes, features = ( data[0].to(self.device), data[1].to(self.device), ) gt_results = data[2] pred_results = self.state.model(segment_sizes, features) segment_sizes, features, pred_results = ( segment_sizes.detach().cpu(), features.detach().cpu(), pred_results.detach().cpu(), ) if gt_results is not None: test_loss_batch = lambda_rank_loss(pred_results, gt_results).item() for k in [1, 5, 10]: test_scores_batch.append(topk_score(pred_results, gt_results, k)) return pred_results.numpy(), test_loss_batch, test_scores_batch def train_full(self): # pylint: disable=too-many-locals """Training on the full dataset.""" # split into training and testing set keys = list(self.state.data.keys()) test_keys = random.sample(keys, k=math.floor(len(keys) * self.test_split)) train_data = OrderedDict() test_data = OrderedDict() for key in keys: if key in test_keys: test_data[key] = self.state.data[key] else: train_data[key] = self.state.data[key] train_features = list( itertools_chain.from_iterable([g.features for g in train_data.values()]) ) test_features = list( itertools_chain.from_iterable([g.features for g in test_data.values()]) ) train_results = np.concatenate([g.min_cost / g.costs for g in train_data.values()]) test_results = np.concatenate([g.min_cost / g.costs for g in test_data.values()]) train_loader = SegmentDataLoader( train_features, train_results, batch_size=self.batch_size, shuffle=True ) test_loader = SegmentDataLoader( test_features, test_results, batch_size=self.batch_size, shuffle=False ) self.optimizer = torch.optim.Adam( self.state.model.parameters(), lr=self.learning_rate, weight_decay=self.weight_decay ) self.scheduler = torch.optim.lr_scheduler.StepLR( self.optimizer, step_size=self.num_epoch_full // 10, gamma=0.8, verbose=True, ) self.state.model = self.state.model.to(self.device) min_test_loss = 1e10 logger.info("Training size: %d; Testing size: %d", len(train_loader), len(test_loader)) model_cache_path = tempfile.NamedTemporaryFile().name # pylint: disable=consider-using-with for epoch in range(self.num_epoch_full): logger.info("Epoch: %d", epoch) # training self.state.model.train() train_loss = None for batch, data in enumerate(train_loader): train_loss = self.train_step(data, batch, train_loss) self.scheduler.step() # testing if epoch % self.test_interval == 0: self.state.model.eval() test_losses, test_scores = [], [] for data in test_loader: _, test_loss_batch, test_scores_batch = self.predict_step(data) test_losses.append(test_loss_batch) test_scores.append(test_scores_batch) test_loss = ( np.array(test_losses[:-1]).mean() if len(test_losses) > 1 else test_losses[0] ) logger.info( "Average test loss: %6f, top1 score: %5f, top5 score: %5f, top10 score: %5f", test_loss, np.array(test_scores)[:, 0].mean(), np.array(test_scores)[:, 1].mean(), np.array(test_scores)[:, 2].mean(), ) if test_loss < min_test_loss: min_test_loss = test_loss torch.save(self.state.model.state_dict(), model_cache_path) self.state.model.to("cpu").load_state_dict(torch.load(model_cache_path)) self.state.untrained_size = 0 def train_incremental( self, features: List[np.ndarray], results: np.ndarray, ): """Training on incremental data. Parameters ---------- features: List[np.ndarray] The extracted features. results: np.ndarray The measured results. """ results = np.min(results) / results loader = SegmentDataLoader(features, results, batch_size=self.batch_size, shuffle=True) self.optimizer = torch.optim.Adam( self.state.model.parameters(), lr=self.learning_rate, weight_decay=self.weight_decay ) self.state.model = self.state.model.to(self.device) logger.info("Incremental training size: %d", len(loader)) for epoch in range(self.num_epoch_incremental): logger.info("Epoch: %d", epoch) self.state.model.train() loss = None for batch, data in enumerate(loader): loss = self.train_step(data, batch, loss) self.state.model.to("cpu") self.state.untrained_size = max(0, self.state.untrained_size - len(loader)) def predict_incremental( self, features: List[np.ndarray], results: Optional[np.ndarray] = None, ) -> np.ndarray: """Predicting (validating) on incremental data. Parameters ---------- features: List[np.ndarray] The extracted features. results: Optional[np.ndarray] The measured results, can be None if used for predicting. Returns ------- pred_results: np.ndarray The predicted results. """ if results is not None: results = np.min(results) / results loader = SegmentDataLoader(features, results, batch_size=self.batch_size, shuffle=False) self.state.model = self.state.model.to(self.device).eval() logger.info("Incremental testing size: %d", len(loader)) pred_results, losses, scores = [], [], [] for data in loader: pred_results_batch, losses_batch, scores_batch = self.predict_step(data) pred_results.append(pred_results_batch) losses.append(losses_batch) scores.append(scores_batch) pred_results = np.concatenate(pred_results) if results is not None: losses = np.array(losses[:-1]).mean() if len(losses) > 1 else losses[0] logger.info( "Average test loss: %6f, top1 score: %5f, top5 score: %5f, top10 score: %5f", losses, np.array(scores)[:, 0].mean(), np.array(scores)[:, 1].mean(), np.array(scores)[:, 2].mean(), ) return pred_results def update( self, features: List[np.ndarray], costs: np.ndarray, group_hash: str, ): """Update the dataset and re-train the model if not frozen. Parameters ---------- features: List[np.ndarray] The extracted features. costs: np.ndarray The measured results. group_hash: str The hash of the group. """ self.state.add_to_group(features, costs, group_hash) if not self.frozen: self.predict_incremental(features, costs) if self.state.untrained_size / self.state.data_size > 0.2: self.train_full() else: self.train_incremental(features, costs) @derived_object class MLPModel(PyCostModel): """Segment Sum MLP Model Parameters ---------- trainer: SegmentSumMLPTrainer The trainer for the model, handling the training interface. """ trainer: SegmentSumMLPTrainer def __init__( self, *, trainer: Optional[SegmentSumMLPTrainer] = None, ): super().__init__() self.trainer = trainer or SegmentSumMLPTrainer() def load(self, path: str) -> None: """Load the cost model, cached data or raw data from given file location. Parameters ---------- path : str The file path. """ self.trainer.state.load(path) def save(self, path: str) -> None: """Save the cost model and data to given file location. Parameters ---------- path : str The file path. """ self.trainer.state.save(path) def update( self, context: TuneContext, candidates: List[MeasureCandidate], results: List[RunnerResult], ) -> None: """Update the dataset, re-train the cost model if not frozen. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. results : List[RunnerResult] The running results of the measure candidates. """ features, mean_costs = extract_features( context, candidates, results, self.trainer.state.extractor ) self.trainer.update(features, mean_costs, shash2hex(context.mod)) def predict(self, context: TuneContext, candidates: List[MeasureCandidate]) -> np.ndarray: """Predict given the measure candidates. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. Return ------ result : np.ndarray The predicted normalized score. """ features, _ = extract_features(context, candidates, None, self.trainer.state.extractor) pred_results = self.trainer.predict_incremental(features) return pred_results
34,492
32.949803
100
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/xgb_model.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """XGBoost-based cost model""" import os import tempfile from collections import OrderedDict from itertools import chain as itertools_chain from typing import TYPE_CHECKING, Any, Callable, Dict, List, NamedTuple, Optional, Tuple import numpy as np # type: ignore from ...contrib.tar import tar, untar from ...runtime import NDArray from ..cost_model import PyCostModel from ..feature_extractor import FeatureExtractor from ..logging import get_logger from ..runner import RunnerResult from ..search_strategy import MeasureCandidate from ..utils import cpu_count, derived_object, shash2hex from .metric import max_curve if TYPE_CHECKING: import xgboost as xgb # type: ignore from xgboost.callback import TrainingCallback # type: ignore from ..tune_context import TuneContext logger = get_logger(__name__) # pylint: disable=invalid-name def make_metric_sorter(focused_metric): """Make sure the focused metric is the first one.""" def metric_name_for_sort(name): if focused_metric == name: return "!" + name return name def sort_key(key): key, _ = key return metric_name_for_sort(key) return sort_key class PackSum: """The pack-sum format Parameters ---------- dmatrix : xgb.DMatrix A float64 array of shape [n, m], where `n` is the packed number of blocks, and `m` is the length of feature vector on each block ids : np.ndarray An int64 array of shape [n] containing nonnegative integers, indicating which the index of a sample that a block belongs to """ dmatrix: "xgb.DMatrix" # type: ignore # pylint: disable=invalid-name ids: np.ndarray def __init__( self, xs: List[np.ndarray], # pylint: disable=invalid-name ys: Optional[np.ndarray], # pylint: disable=invalid-name ): """Create PackSum format given a batch of samples Parameters ---------- xs : List[np.ndarray] A batch of input samples ys : Optional[List[float]] A batch of labels. None means no labels available. """ import xgboost as xgb # type: ignore # pylint: disable=import-outside-toplevel repeats = [x.shape[0] for x in xs] xs = np.concatenate(xs, axis=0) self.ids = np.concatenate([[i] * repeat for i, repeat in enumerate(repeats)], axis=0) if ys is None: self.dmatrix = xgb.DMatrix(data=xs, label=None) else: ys = np.concatenate([[y] * repeat for y, repeat in zip(ys, repeats)], axis=0) self.dmatrix = xgb.DMatrix(data=xs, label=ys) self.dmatrix.set_weight(ys) def predict_with_score(self, pred: np.ndarray) -> np.ndarray: """Predict the labels given the block level prediction scores. Parameters ---------- pred : np.ndarray The block level predictions Returns ------- result : np.ndarray The predictions for each candidate. """ return np.bincount(self.ids, weights=pred) def obj_square_error(self, ys_pred: np.ndarray) -> Tuple[np.ndarray, np.ndarray]: """Implement square error loss on pack-sum format as a custom objective function for xgboost. Parameters ---------- ys_pred: np.ndarray The predictions Returns ------- gradient: np.ndarray The gradient according to the xgboost format hessian: np.ndarray The hessian according to the xgboost format """ # Making prediction ys_pred = self.predict_with_score(ys_pred) # Propagate prediction to each block ys_pred = ys_pred[self.ids] # pylint: disable=invalid-sequence-index # The gradient and hessian ys = self.dmatrix.get_label() # type: ignore # pylint: disable=invalid-name gradient = ys_pred - ys hessian = np.ones_like(gradient) return gradient * ys, hessian * ys def rmse(self, ys_pred: np.ndarray) -> Tuple[str, float]: """Evaluate RMSE (rooted mean square error) in the pack-sum format Parameters ---------- ys_pred: np.ndarray The raw predictions Returns ------- name: str The name of the metric score: float The score of the metric """ # Making prediction ys_pred = self.predict_with_score(ys_pred) # Propagate prediction to each block ys_pred = ys_pred[self.ids] # pylint: disable=invalid-sequence-index # The RMSE ys = self.dmatrix.get_label() # type: ignore # pylint: disable=invalid-name square_error = np.square(ys_pred - ys) rmse = np.sqrt(square_error.mean()) return "p-rmse", rmse def average_peak_score( self, ys_pred: np.ndarray, n: int, ) -> Tuple[str, float]: """Evaluate average-peak-score@N in the pack-sum format Parameters ---------- ys_pred: np.ndarray The raw prediction n : int The N in average-peak-score@N Returns ------- name: str The name of the metric score: float The score of the metric """ ys = self.dmatrix.get_label() # type: ignore # pylint: disable=invalid-name ys = self.predict_with_score(ys) # type: ignore # pylint: disable=invalid-name ys = ys / np.unique(self.ids, return_counts=True)[1] # type: ignore # pylint: disable=invalid-name ys_pred = self.predict_with_score(ys_pred) trials = np.argsort(ys_pred)[::-1][:n] trial_scores = ys[trials] curve = max_curve(trial_scores) / np.max(ys) score = np.mean(curve) return f"a-peak@{n}", score class XGBConfig(NamedTuple): """XGBoost model configuration Parameters ---------- max_depth : int The maximum depth. gamma : float The gamma. min_child_weight : float The minimum child weight. eta : float The eta, learning rate. seed : int The random seed. nthread : Optional[int], The number of threads to use. Default is None, which means to use physical number of cores. """ max_depth: int = 10 gamma: float = 0.001 min_child_weight: float = 0 eta: float = 0.2 seed: int = 43 nthread: Optional[int] = None def to_dict(self): return { "max_depth": self.max_depth, "gamma": self.gamma, "min_child_weight": self.min_child_weight, "eta": self.eta, "seed": self.seed, "nthread": self.nthread, } class FeatureGroup: """Feature group Parameters ---------- group_hash : str The hash of the group features : List[np.ndarray] The features costs : List[float] The costs min_cost : float The minimum cost """ group_hash: str features: List[np.ndarray] costs: np.ndarray min_cost: float def __init__( self, group_hash: str, features: List[np.ndarray], costs: np.ndarray, ) -> None: self.group_hash = group_hash self.features = features self.costs = costs self.min_cost = np.min(costs) def append( self, features: List[np.ndarray], costs: np.ndarray, ) -> None: self.features.extend(features) self.costs = np.append(self.costs, costs) self.min_cost = np.min(self.costs) @derived_object class XGBModel(PyCostModel): """XGBoost model Parameters ---------- extractor : FeatureExtractor The feature extractor for the model. config : XGBConfig The XGBoost model config. num_warmup_samples : int The number of samples that are used for warmup, i.e., the first few samples are predicted with random results. early_stopping_rounds : int The number of rounds for early stopping. verbose_eval : int The verbose level when doing evaluation. average_peak_n : int The number to calculate average peak score. adaptive_training : bool Whether use adaptive training to reduce tuning time. """ # feature extractor extractor: FeatureExtractor # xgboost model config config: XGBConfig # behavior of randomness num_warmup_samples: int # evaluation early_stopping_rounds: int verbose_eval: int average_peak_n: int # states data: Dict[str, FeatureGroup] data_size: int booster: Optional["xgb.Booster"] # adaptive training adaptive_training: bool last_train_size: int def __init__( self, *, # feature extractor extractor: FeatureExtractor.FeatureExtractorType = "per-store-feature", # xgboost model config config: XGBConfig = XGBConfig(), # random result before enough samples num_warmup_samples: int = 100, # evaluation early_stopping_rounds: int = 50, verbose_eval: int = 25, average_peak_n: int = 32, adaptive_training: bool = True, num_tuning_cores: Optional[int] = None, ): super().__init__() if not isinstance(extractor, FeatureExtractor): extractor = FeatureExtractor.create(extractor) # feature extractor self.extractor = extractor # model-related if config.nthread is None: # use physical core number if num_tuning_cores is None: config = config._replace(nthread=cpu_count(logical=False)) else: config = config._replace(nthread=num_tuning_cores) self.config = config # behavior of randomness self.num_warmup_samples = num_warmup_samples # evaluation self.early_stopping_rounds = early_stopping_rounds self.verbose_eval = verbose_eval self.average_peak_n = average_peak_n # states self.data = OrderedDict() self.data_size = 0 self.booster = None # adaptive training self.adaptive_training = adaptive_training self.last_train_size = 0 def load(self, path: str) -> None: """Load the cost model from given file location. Parameters ---------- path : str The file path. Note ---- Since XGBoost model trains from scratch, each time this method loads the model together with previously cached feature vectors and results, so that the subsequent training process could use all the existing data being stored on disk. """ import xgboost as xgb # pylint: disable=import-outside-toplevel with tempfile.TemporaryDirectory() as tmp_dir: model_path = os.path.join(tmp_dir, "model.bin") data_path = os.path.join(tmp_dir, "data.npy") # Step 1. Untar untar(path, tmp_dir) # Step 2. Load data data = OrderedDict() data_size = 0 for group_hash, features, costs in np.load(data_path, allow_pickle=True): data[group_hash] = FeatureGroup( group_hash=group_hash, features=list(features), costs=costs, ) data_size += len(costs) # Step 3. Load the model if os.path.exists(model_path): booster = xgb.Booster() booster.load_model(model_path) else: self.booster = None self.data = data self.data_size = data_size self.booster = booster def save(self, path: str) -> None: """Save the cost model to given file location. Parameters ---------- path : str The file path. Note ---- Since XGBoost model trains from scratch, each time this method saves the model together with previously cached feature vectors and results, so that the subsequent training process could use all the existing data being stored on disk. """ with tempfile.TemporaryDirectory() as tmp_dir: model_path = os.path.join(tmp_dir, "model.bin") data_path = os.path.join(tmp_dir, "data.npy") # Step 1. Save the model booster = self.booster if booster is not None: booster.save_model(model_path) else: model_path = None # Step 2. Save data data = [ ( g.group_hash, g.features, g.costs, ) for g in self.data.values() ] np.save( file=data_path, arr=np.array(data, dtype=object), ) # Step 3. Tar it tar(path, [x for x in [model_path, data_path] if x is not None]) logger.info("Saved XGBModel to %s", path) def update( self, context: "TuneContext", candidates: List[MeasureCandidate], results: List[RunnerResult], ) -> None: """Update the cost model given running results. Parameters ---------- context : TuneContext The tuning context. candidates : List[MeasureCandidate] The measure candidates. results : List[RunnerResult] The running results of the measure candidates. """ assert len(candidates) == len(results) if len(candidates) == 0: return # Step 1. Get the feature group new_group_hash = shash2hex(context.mod) group = self.data.get(new_group_hash, None) # Step 2. Extract features def _feature(x: NDArray) -> np.ndarray: return x.numpy().astype("float32") def _mean_cost(x: RunnerResult) -> float: if not x.run_secs: return 1e10 return float(np.median([float(s) for s in x.run_secs])) new_features = [_feature(x) for x in self.extractor.extract_from(context, candidates)] new_mean_costs = [_mean_cost(x) for x in results] # Filter instances with no features new_mean_costs = [c for i, c in enumerate(new_mean_costs) if len(new_features[i]) != 0] new_mean_costs_np = np.array(new_mean_costs).astype("float32") new_features = [f for f in new_features if len(f) != 0] if not new_features: return # Steps 3. Run validation if group is not None and self.booster is not None: logger.debug( "XGB validation: %s", "\t".join( f"{key}: {score:.6f}" for key, score in self._validate( xs=new_features, ys=group.min_cost / new_mean_costs_np, ) ), ) # Step 4. Add the features into the data points if group is None: group = FeatureGroup( group_hash=new_group_hash, features=new_features, costs=new_mean_costs_np, ) else: group.append(new_features, new_mean_costs_np) self.data[new_group_hash] = group self.data_size += len(new_features) if ( self.adaptive_training and self.data_size - self.last_train_size < self.last_train_size / 5 ): # Set a training threshold related to `last_train_size` to reduce the training # overhead when there're too many results return self.last_train_size = self.data_size # Step 5. Re-train the model self._train( xs=list(itertools_chain.from_iterable([g.features for g in self.data.values()])), ys=np.concatenate( [g.min_cost / g.costs for g in self.data.values()], axis=0, ), ) def predict( self, context: "TuneContext", candidates: List[MeasureCandidate], ) -> np.ndarray: """Predict the normalized score using the cost model. Parameters ---------- context : TuneContext The tuning context. candidates : List[MeasureCandidate] The measure candidates. Return ------ result : np.ndarray The predicted normalized score. """ if self.data_size >= self.num_warmup_samples and self.booster is not None: ret = self._predict( xs=[ x.numpy().astype("float32") for x in self.extractor.extract_from( context, candidates, ) ] ) else: ret = np.random.uniform( low=0, high=1, size=(len(candidates),), ) return ret.astype("float64") def _train( # type: ignore # pylint: disable=invalid-name self, xs: List[np.ndarray], ys: np.ndarray, ) -> None: import xgboost as xgb # type: ignore # pylint: disable=import-outside-toplevel self.d_train = PackSum(xs=xs, ys=ys) def obj(ys_pred: np.ndarray, d_train: "xgb.DMatrix"): # type: ignore # pylint: disable = unused-argument return self.d_train.obj_square_error(ys_pred) def rmse(ys_pred: np.ndarray, d_train: "xgb.DMatrix"): # type: ignore # pylint: disable = unused-argument return self.d_train.rmse(ys_pred) def avg_peak_score(ys_pred: np.ndarray, d_train: "xgb.DMatrix"): # type: ignore # pylint: disable = unused-argument return self.d_train.average_peak_score(ys_pred, self.average_peak_n) self.booster = xgb.train( self.config.to_dict(), self.d_train.dmatrix, num_boost_round=10000, obj=obj, callbacks=[ _get_custom_call_back( early_stopping_rounds=self.early_stopping_rounds, verbose_eval=self.verbose_eval, fevals=[rmse, avg_peak_score], evals=[(self.d_train.dmatrix, "tr")], cvfolds=None, ) ], ) del self.d_train def _predict( # type: ignore # pylint: disable=invalid-name self, xs: List[np.ndarray], ) -> np.ndarray: d_test = PackSum(xs=xs, ys=None) pred = self.booster.predict(d_test.dmatrix) ret = d_test.predict_with_score(pred) return ret def _validate( # type: ignore # pylint: disable=invalid-name self, xs: List[np.ndarray], ys: np.ndarray, ) -> List[Tuple[str, float]]: """Evaluate the score of inputs. Parameters ---------- xs : List[np.ndarray] A batch of input samples ys : List[float] A batch of labels Returns ------- scores: np.ndarray The predicted result for all inputs. """ assert self.booster is not None d_valid = PackSum(xs=xs, ys=ys) def average_peak_score(ys_pred: np.ndarray): return d_valid.average_peak_score(ys_pred, n=self.average_peak_n) ys_pred = self.booster.predict(d_valid.dmatrix) eval_result: List[Tuple[str, float]] = [ feval(ys_pred) for feval in ( average_peak_score, d_valid.rmse, ) ] eval_result.sort(key=make_metric_sorter("p-rmse")) return eval_result def _get_custom_call_back( early_stopping_rounds: int, verbose_eval: int, fevals: List[Callable], evals: List[Tuple["xgb.DMatrix", str]], focused_metric: str = "tr-p-rmse", cvfolds: List["xgb.training.CVPack"] = None, ) -> "TrainingCallback": """Get a customized callback function for XGBoost. Work around xgboost import.""" def optional_xgboost_callback(cls): """Decorator for importing TrainingCallback from xgboost""" # pylint:disable = import-outside-toplevel try: from xgboost.callback import TrainingCallback # type: ignore # pylint:enable = import-outside-toplevel except ImportError: class TrainingCallback: # type: ignore pass class OptXGBoostCustomCallback(cls, TrainingCallback): # type: ignore pass return OptXGBoostCustomCallback @optional_xgboost_callback class XGBoostCustomCallback: """Custom callback class for xgboost to support multiple custom evaluation functions""" def __init__( self, early_stopping_rounds: int, verbose_eval: int, fevals: List[Callable], evals: List[Tuple["xgb.DMatrix", str]], focused_metric: str = "tr-p-rmse", cvfolds: List["xgb.training.CVPack"] = None, ): self.early_stopping_rounds = early_stopping_rounds self.verbose_eval = verbose_eval self.fevals = fevals self.evals = evals self.state: Dict[str, Any] = {} self.focused_metric = focused_metric self.sort_key = make_metric_sorter(focused_metric=focused_metric) self.cvfolds = cvfolds if cvfolds is not None: self.aggregated_cv = None def __call__(self, env: "xgb.core.CallbackEnv"): # Compatibility with xgboost < 1.3 return self.after_iteration(env.model, env.iteration, env.evaluation_result_list) def init(self, model: "xgb.Booster"): """Internal function for initialization""" booster: "xgb.Booster" = model self.state["best_iteration"] = 0 self.state["best_score"] = float("inf") if booster is None: assert self.cvfolds is not None return if booster.attr("best_score") is not None: self.state["best_score"] = float(booster.attr("best_score")) self.state["best_iteration"] = int(booster.attr("best_iteration")) self.state["best_msg"] = booster.attr("best_msg") else: booster.set_attr(best_iteration=str(self.state["best_iteration"])) booster.set_attr(best_score=str(self.state["best_score"])) def after_iteration( self, model: "xgb.Booster", epoch: int, evals_log: Dict ): # pylint: disable = unused-argument """Internal function for after_iteration""" # pylint:disable = import-outside-toplevel try: from xgboost.callback import _fmt_metric # type: ignore except ImportError: # Compatibility with xgboost >= 1.6 def _fmt_metric(value, show_stdv=True): if len(value) == 2: return f"{value[0]}:{value[1]:.5f}" if len(value) == 3: if show_stdv: return f"{value[0]}:{value[1]:.5f}+{value[2]:.5f}" return f"{value[0]}:{value[1]:.5f}" raise ValueError("wrong metric value", value) import xgboost as xgb from xgboost import rabit # type: ignore try: from xgboost.training import aggcv # type: ignore except ImportError: from xgboost.callback import _aggcv as aggcv # type: ignore # pylint:enable = import-outside-toplevel if not self.state: self.init(model) booster: xgb.Booster = model iteration: int = epoch cvfolds: List[xgb.training.CVPack] = self.cvfolds ##### Evaluation ##### # `eval_result` is a list of (key, score) eval_result: List[Tuple[str, float]] = [] if cvfolds is None: eval_result = list( itertools_chain.from_iterable( [ (key, float(value)) for key, value in map( lambda x: x.split(":"), booster.eval_set( evals=self.evals, iteration=iteration, feval=feval, ).split()[1:], ) ] for feval in self.fevals ) ) else: eval_result = list( itertools_chain.from_iterable( [ (key, score) for key, score, _std in aggcv( fold.eval( iteration=iteration, feval=feval, ) for fold in cvfolds ) ] for feval in self.fevals ) ) eval_result = list(eval_result) eval_result.sort(key=self.sort_key) ##### Print eval result ##### if self.verbose_eval and iteration % self.verbose_eval == 0: info = [] for key, score in eval_result: if "null" not in key: info.append(f"{key}: {score:.6f}") logger.debug("XGB iter %3d: %s", iteration, "\t".join(info)) ##### Choose score and do early stopping ##### score = None for key, _score in eval_result: if key == self.focused_metric: score = _score break assert score is not None best_score = self.state["best_score"] best_iteration = self.state["best_iteration"] if score < best_score: tab = "\t" # to work with f-string msg = f"[{epoch}] {tab.join([_fmt_metric(x) for x in eval_result])}" self.state["best_msg"] = msg self.state["best_score"] = score self.state["best_iteration"] = epoch # save the property to attributes, so they will occur in checkpoint. if model is not None: model.set_attr( best_score=str(self.state["best_score"]), best_iteration=str(self.state["best_iteration"]), best_msg=self.state["best_msg"], ) elif epoch - best_iteration >= self.early_stopping_rounds: best_msg = self.state["best_msg"] if self.verbose_eval and rabit.get_rank() == 0: logger.debug("XGB stopped. Best iteration: %s ", best_msg) # instead of raising EarlyStopException, returning True to end the training return True # False to indicate training should not stop. return False return XGBoostCustomCallback( early_stopping_rounds=early_stopping_rounds, verbose_eval=verbose_eval, fevals=fevals, evals=evals, focused_metric=focused_metric, cvfolds=cvfolds, )
28,769
33.047337
124
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/random_model.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ Random cost model """ from typing import List, Optional, Tuple, Union from ..cost_model import PyCostModel from ..runner import RunnerResult from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext from ..utils import derived_object # type: ignore @derived_object class RandomModel(PyCostModel): """Random cost model Parameters ---------- random_state : Union[Tuple[str, np.ndarray, int, int, float], dict] The random state of the random number generator. path : Optional[str] The path of the random cost model. max_range : Optional[int] The maximum range of random results, [0, max_range]. Reference --------- https://numpy.org/doc/stable/reference/random/generated/numpy.random.get_state.html """ import numpy as np # type: ignore # pylint: disable=import-outside-toplevel random_state: Union[Tuple[str, np.ndarray, int, int, float], dict] path: Optional[str] def __init__( self, *, seed: Optional[int] = None, path: Optional[str] = None, max_range: Optional[int] = 100, ): import numpy as np # type: ignore # pylint: disable=import-outside-toplevel super().__init__() if path is not None: self.load(path) else: np.random.seed(seed) self.random_state = np.random.get_state() self.max_range = max_range def load(self, path: str) -> None: """Load the cost model from given file location. Parameters ---------- path : str The file path. """ import numpy as np # type: ignore # pylint: disable=import-outside-toplevel self.random_state = tuple(np.load(path, allow_pickle=True)) # type: ignore def save(self, path: str) -> None: """Save the cost model to given file location. Parameters ---------- path : str The file path. """ import numpy as np # type: ignore # pylint: disable=import-outside-toplevel np.save(path, np.array(self.random_state, dtype=object), allow_pickle=True) def update( self, context: TuneContext, candidates: List[MeasureCandidate], results: List[RunnerResult], ) -> None: """Update the cost model given running results. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. results : List[RunnerResult] The running results of the measure candidates. """ def predict( self, context: TuneContext, candidates: List[MeasureCandidate] ) -> np.ndarray: # type: ignore # pylint: disable=used-before-assignment """Update the cost model given running results. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. Return ------ result : np.ndarray The predicted running results. """ import numpy as np # type: ignore # pylint: disable=import-outside-toplevel np.random.set_state(self.random_state) # TODO(@zxybazh): Use numpy's RandState object: # https://numpy.org/doc/1.16/reference/generated/numpy.random.RandomState.html#numpy.random.RandomState result = np.random.rand(len(candidates)) * self.max_range # type: ignore self.random_state = np.random.get_state() return result
4,437
31.632353
111
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.cost_model package. """ from .cost_model import CostModel, PyCostModel from .random_model import RandomModel from .xgb_model import XGBModel
952
40.434783
62
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/cost_model.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Meta Schedule CostModel.""" import ctypes from typing import Callable, List, Union # isort: off from typing_extensions import Literal # isort: on import numpy as np # type: ignore from tvm._ffi import register_object from tvm.runtime import Object from .. import _ffi_api from ..runner import RunnerResult from ..search_strategy import MeasureCandidate from ..tune_context import TuneContext from ..utils import _get_default_str @register_object("meta_schedule.CostModel") class CostModel(Object): """Cost model.""" CostModelType = Union["CostModel", Literal["xgb", "mlp", "random"]] def load(self, path: str) -> None: """Load the cost model from given file location. Parameters ---------- path : str The file path. """ _ffi_api.CostModelLoad(self, path) # type: ignore # pylint: disable=no-member def save(self, path: str) -> None: """Save the cost model to given file location. Parameters ---------- path : str The file path. """ _ffi_api.CostModelSave(self, path) # type: ignore # pylint: disable=no-member def update( self, context: TuneContext, candidates: List[MeasureCandidate], results: List[RunnerResult], ) -> None: """Update the cost model given running results. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. results : List[RunnerResult] The running results of the measure candidates. """ _ffi_api.CostModelUpdate(self, context, candidates, results) # type: ignore # pylint: disable=no-member def predict(self, context: TuneContext, candidates: List[MeasureCandidate]) -> np.ndarray: """Predict normalized score with the cost model. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. Return ------ result : np.ndarray The predicted normalized score. """ n = len(candidates) results = np.zeros(shape=(n,), dtype="float64") _ffi_api.CostModelPredict( # type: ignore # pylint: disable=no-member self, context, candidates, results.ctypes.data_as(ctypes.c_void_p), ) return results @staticmethod def create( kind: Literal["xgb", "mlp", "random", "none"], *args, **kwargs, ) -> "CostModel": """Create a CostModel. Parameters ---------- kind : Literal["xgb", "mlp", "random", "none"] The kind of the cost model. Can be "xgb", "mlp", "random" or "none". Returns ------- cost_model : CostModel The created cost model. """ from . import RandomModel, XGBModel # pylint: disable=import-outside-toplevel if kind == "xgb": return XGBModel(*args, **kwargs) # type: ignore if "num_tuning_cores" in kwargs: # num_tuning_cores is only relevant for XGBModel. kwargs.pop("num_tuning_cores") if kind == "random": return RandomModel(*args, **kwargs) # type: ignore if kind == "mlp": from .mlp_model import ( # type: ignore # pylint: disable=import-outside-toplevel MLPModel, ) return MLPModel(*args, **kwargs) # type: ignore if kind == "none": return None # no cost model required raise ValueError(f"Unknown CostModel: {kind}") create = CostModel.create # pylint: disable=invalid-name @register_object("meta_schedule.PyCostModel") class _PyCostModel(CostModel): """ A TVM object cost model to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PyCostModel """ def __init__( self, f_load: Callable = None, f_save: Callable = None, f_update: Callable = None, predict_func: Callable = None, f_as_string: Callable = None, ): """Constructor.""" def f_predict(context: TuneContext, candidates: List[MeasureCandidate], return_ptr) -> None: n = len(candidates) return_ptr = ctypes.cast(return_ptr, ctypes.POINTER(ctypes.c_double)) array_wrapper = np.ctypeslib.as_array(return_ptr, shape=(n,)) res = predict_func(context, candidates) array_wrapper[:] = res assert ( array_wrapper.dtype == "float64" ), "ValueError: Invalid data type returned from CostModel Predict!" self.__init_handle_by_constructor__( _ffi_api.CostModelPyCostModel, # type: ignore # pylint: disable=no-member f_load, f_save, f_update, f_predict, f_as_string, ) class PyCostModel: """ An abstract cost model with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PyCostModel, "methods": ["load", "save", "update", "predict", "__str__"], } def load(self, path: str) -> None: """Load the cost model from given file location. Parameters ---------- path : str The file path. """ raise NotImplementedError def save(self, path: str) -> None: """Save the cost model to given file location. Parameters ---------- path : str The file path. """ raise NotImplementedError def update( self, context: TuneContext, candidates: List[MeasureCandidate], results: List[RunnerResult], ) -> None: """Update the cost model given running results. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. results : List[RunnerResult] The running results of the measure candidates. """ raise NotImplementedError def predict(self, context: TuneContext, candidates: List[MeasureCandidate]) -> np.ndarray: """Predict given the measure candidates. Parameters ---------- context : TuneContext, The tuning context. candidates : List[MeasureCandidate] The measure candidates. Return ------ result : np.ndarray The predicted normalized score. """ raise NotImplementedError def __str__(self) -> str: """Get the cost model as string with name. Return ------ result : str Get the cost model as string with name. """ return _get_default_str(self)
8,007
28.992509
112
py
tvm
tvm-main/python/tvm/meta_schedule/cost_model/metric.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Cost model metrics for meta schedule""" import numpy as np # type: ignore def max_curve(trial_scores: np.ndarray) -> np.ndarray: """f(n) = max([s[i] fo i < n]) Parameters ---------- trial_scores : List[float] the score of i-th trial Returns ------- curve : np.ndarray A vector, the max-curve function values """ ret = np.empty(len(trial_scores)) keep = -1e9 for i, score in enumerate(trial_scores): keep = max(keep, score) ret[i] = keep return ret
1,321
32.05
62
py
tvm
tvm-main/python/tvm/meta_schedule/space_generator/post_order_apply.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Post Order Apply Space Generator.""" from tvm._ffi import register_object from .. import _ffi_api from .space_generator import ( MutatorProbType, PostprocType, ScheduleRuleType, SpaceGenerator, _normalize_rules, ) @register_object("meta_schedule.PostOrderApply") class PostOrderApply(SpaceGenerator): """ PostOrderApply is the design space generator that generates design spaces by applying schedule rules to blocks in post-DFS order. Parameters ---------- f_block_filter : Optional[function] An optional callback function that is used to filter which blocks have schedules generated for them. The function should take in a block and return True if a schedule should be generated or False if that block should be skipped. If no function is provided all blocks will have schedules generated. """ def __init__( self, f_block_filter=None, sch_rules: ScheduleRuleType = "from-target", postprocs: PostprocType = "from-target", mutator_probs: MutatorProbType = "from-target", ): """Constructor""" sch_rules, postprocs, mutator_probs = _normalize_rules(sch_rules, postprocs, mutator_probs) self.__init_handle_by_constructor__( _ffi_api.SpaceGeneratorPostOrderApply, # type: ignore # pylint: disable=no-member f_block_filter, sch_rules, postprocs, mutator_probs, )
2,271
36.245902
99
py
tvm
tvm-main/python/tvm/meta_schedule/space_generator/space_generator_union.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Union of meta Schedule design space generators.""" from typing import List from tvm._ffi import register_object from .. import _ffi_api from .space_generator import ( MutatorProbType, PostprocType, ScheduleRuleType, SpaceGenerator, _normalize_rules, ) @register_object("meta_schedule.SpaceGeneratorUnion") class SpaceGeneratorUnion(SpaceGenerator): """Union of design space generators.""" def __init__( self, space_generators: List[SpaceGenerator], sch_rules: ScheduleRuleType = "from-target", postprocs: PostprocType = "from-target", mutator_probs: MutatorProbType = "from-target", ): """Constructor. Parameters ---------- space_generators : List[SpaceGenerator] The list of design space generators to be unioned. """ sch_rules, postprocs, mutator_probs = _normalize_rules(sch_rules, postprocs, mutator_probs) self.__init_handle_by_constructor__( _ffi_api.SpaceGeneratorSpaceGeneratorUnion, # type: ignore # pylint: disable=no-member space_generators, sch_rules, postprocs, mutator_probs, )
1,995
33.413793
99
py
tvm
tvm-main/python/tvm/meta_schedule/space_generator/space_generator.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ Meta Schedule design space generators that generates design space for generation of measure candidates. """ from typing import TYPE_CHECKING, Callable, Dict, List, Optional, Tuple, Union # isort: off from typing_extensions import Literal # isort: on from tvm._ffi import register_object from tvm.ir import IRModule from tvm.runtime import Object from tvm.tir.schedule import Schedule from .. import _ffi_api if TYPE_CHECKING: from ..mutator import Mutator from ..postproc import Postproc from ..schedule_rule import ScheduleRule from ..tune_context import TuneContext @register_object("meta_schedule.SpaceGenerator") class SpaceGenerator(Object): """The abstract design space generator interface.""" ScheduleFnType = Union[ Callable[[Schedule], None], # No output Callable[[Schedule], Schedule], # Single output Callable[[Schedule], List[Schedule]], # Multiple outputs ] SpaceGeneratorType = Union[ "SpaceGenerator", ScheduleFnType, Literal["post-order-apply", "union"], ] sch_rules: Optional[List["ScheduleRule"]] postprocs: Optional[List["Postproc"]] mutator_probs: Optional[Dict["Mutator", float]] def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the design space generator with tuning context. Parameters ---------- context : TuneContext The tuning context for initializing the design space generator. """ _ffi_api.SpaceGeneratorInitializeWithTuneContext( # type: ignore # pylint: disable=no-member self, context ) def generate_design_space(self, mod: IRModule) -> List[Schedule]: """Generate design spaces given a module. Parameters ---------- mod : IRModule The module used for design space generation. Returns ------- design_spaces : List[tvm.tir.Schedule] The generated design spaces, i.e., schedules. """ return _ffi_api.SpaceGeneratorGenerateDesignSpace(self, mod) # type: ignore # pylint: disable=no-member def clone(self) -> "SpaceGenerator": """Clone the design space generator. Returns ------- cloned_sg : SpaceGenerator The cloned design space generator. """ return _ffi_api.SpaceGeneratorClone(self) # type: ignore # pylint: disable=no-member @staticmethod def create( # pylint: disable=keyword-arg-before-vararg kind: Union[ Literal["post-order-apply", "union"], ScheduleFnType, ] = "post-order-apply", *args, **kwargs, ) -> "SpaceGenerator": """Create a design space generator.""" from . import ( # pylint: disable=import-outside-toplevel PostOrderApply, ScheduleFn, SpaceGeneratorUnion, ) if callable(kind): def create_schedule_fn( func, sch_rules=[], postprocs=[], mutator_probs={}, ): # pylint: disable=dangerous-default-value return ScheduleFn(func, sch_rules, postprocs, mutator_probs) return create_schedule_fn(kind, *args, **kwargs) # type: ignore if kind == "post-order-apply": return PostOrderApply(*args, **kwargs) if kind == "union": return SpaceGeneratorUnion(*args, **kwargs) if isinstance(kind, str): return PostOrderApply(sch_rules=kind, postprocs=kind, mutator_probs=kind) raise ValueError(f"Unknown SpaceGenerator: {kind}") ScheduleFnType = SpaceGenerator.ScheduleFnType ScheduleRuleType = Union[ List["ScheduleRule"], Literal["llvm", "cuda", "cuda-tensorcore", "hexagon", "from-target"], ] PostprocType = Union[ List["Postproc"], Literal["llvm", "cuda", "cuda-tensorcore", "hexagon", "from-target"], ] MutatorProbType = Union[ Dict["Mutator", float], Literal["llvm", "cuda", "cuda-tensorcore", "hexagon", "from-target"], ] create = SpaceGenerator.create # pylint: disable=invalid-name def _normalize_rules( sch_rules: ScheduleRuleType, postprocs: PostprocType, mutator_probs: MutatorProbType, ) -> Tuple[ Optional[List["ScheduleRule"]], Optional[List["Postproc"]], Optional[Dict["Mutator", float]], ]: # pylint: disable=import-outside-toplevel from ..mutator import Mutator from ..postproc import Postproc from ..schedule_rule import ScheduleRule # pylint: enable=import-outside-toplevel assert sch_rules is not None assert postprocs is not None assert mutator_probs is not None if isinstance(sch_rules, str): if sch_rules == "from-target": sch_rules = None else: sch_rules = ScheduleRule.create(sch_rules) if isinstance(postprocs, str): if postprocs == "from-target": postprocs = None else: postprocs = Postproc.create(postprocs) if isinstance(mutator_probs, str): if mutator_probs == "from-target": mutator_probs = None else: mutator_probs = Mutator.create(mutator_probs) return sch_rules, postprocs, mutator_probs # type: ignore @register_object("meta_schedule.PySpaceGenerator") class _PySpaceGenerator(SpaceGenerator): """ A TVM object space generator to support customization on the python side. This is NOT the user facing class for function overloading inheritance. See also: PySpaceGenerator """ def __init__( self, sch_rules: ScheduleRuleType = "from-target", postprocs: PostprocType = "from-target", mutator_probs: MutatorProbType = "from-target", f_initialize_with_tune_context: Optional[Callable] = None, f_generate_design_space: Optional[Callable] = None, f_clone: Optional[Callable] = None, ): """Constructor.""" sch_rules, postprocs, mutator_probs = _normalize_rules(sch_rules, postprocs, mutator_probs) self.__init_handle_by_constructor__( _ffi_api.SpaceGeneratorPySpaceGenerator, # type: ignore # pylint: disable=no-member sch_rules, postprocs, mutator_probs, f_initialize_with_tune_context, f_generate_design_space, f_clone, ) class PySpaceGenerator: """ An abstract space generator with customized methods on the python-side. This is the user facing class for function overloading inheritance. Note: @derived_object is required for proper usage of any inherited class. """ _tvm_metadata = { "cls": _PySpaceGenerator, "fields": ["sch_rules", "postprocs", "mutator_probs"], "methods": ["_initialize_with_tune_context", "generate_design_space", "clone"], } def _initialize_with_tune_context(self, context: "TuneContext") -> None: """Initialize the design space generator with tuning context. Parameters ---------- context : TuneContext The tuning context for initializing the design space generator. """ raise NotImplementedError def generate_design_space(self, mod: IRModule) -> List[Schedule]: """Generate design spaces given a module. Parameters ---------- mod : IRModule The module used for design space generation. Returns ------- design_spaces : List[tvm.tir.Schedule] The generated design spaces, i.e., schedules. """ raise NotImplementedError def clone(self) -> SpaceGenerator: """Clone the design space generator. Returns ------- cloned_sg : SpaceGenerator The cloned design space generator. """ raise NotImplementedError
8,726
31.685393
112
py
tvm
tvm-main/python/tvm/meta_schedule/space_generator/schedule_fn.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Union of meta Schedule design space generators.""" from tvm._ffi import register_object from .. import _ffi_api from .space_generator import ( MutatorProbType, PostprocType, ScheduleRuleType, SpaceGenerator, _normalize_rules, ) @register_object("meta_schedule.ScheduleFn") class ScheduleFn(SpaceGenerator): """Create a design space generator with customized schedule function. The schedule function can have the following signatures: - 1) [Schedule] -> None - 2) [Schedule] -> Schedule - 3) [Schedule] -> List[Schedule] """ def __init__( self, sch_fn: SpaceGenerator.ScheduleFnType, sch_rules: ScheduleRuleType = "from-target", postprocs: PostprocType = "from-target", mutator_probs: MutatorProbType = "from-target", ): """Constructor. Parameters ---------- sch_fn : SpaceGenerator.ScheduleFnType The schedule function, which can have the following signatures: - 1) [Schedule] -> None - 2) [Schedule] -> Schedule - 3) [Schedule] -> List[Schedule] """ sch_rules, postprocs, mutator_probs = _normalize_rules(sch_rules, postprocs, mutator_probs) self.__init_handle_by_constructor__( _ffi_api.SpaceGeneratorScheduleFn, # type: ignore # pylint: disable=no-member sch_fn, sch_rules, postprocs, mutator_probs, )
2,263
34.375
99
py
tvm
tvm-main/python/tvm/meta_schedule/space_generator/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The tvm.meta_schedule.space_generator package. Meta Schedule design space generators that generates design space for generation of measure candidates. """ from .post_order_apply import PostOrderApply from .schedule_fn import ScheduleFn from .space_generator import PySpaceGenerator, ScheduleFnType, SpaceGenerator, create from .space_generator_union import SpaceGeneratorUnion from ...target import x86
1,193
41.642857
85
py
tvm
tvm-main/python/tvm/target/target.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Target data structure.""" import json import re import warnings import tvm._ffi from tvm._ffi import register_func as _register_func from tvm.runtime import Object, convert from tvm.runtime.container import String from tvm.ir.container import Map, Array from . import _ffi_api @tvm._ffi.register_object class TargetKind(Object): """Kind of a compilation target""" @property def options(self): """Returns the dict of available option names and types""" return dict(_ffi_api.ListTargetKindOptions(self)) @staticmethod def options_from_name(kind_name: str): """Returns the dict of available option names and types from a name of TargetKind""" return dict(_ffi_api.ListTargetKindOptionsFromName(kind_name)) class TargetFeatures: def __init__(self, target): self.target = target def __getattr__(self, name: str): return _ffi_api.TargetGetFeature(self.target, name) @tvm._ffi.register_object class Target(Object): """Target device information, use through TVM API. Note ---- You can create target using the constructor or the following functions - :py:func:`tvm.target.arm_cpu` create arm_cpu target - :py:func:`tvm.target.cuda` create CUDA target - :py:func:`tvm.target.rocm` create ROCM target - :py:func:`tvm.target.mali` create Mali target - :py:func:`tvm.target.intel_graphics` create Intel Graphics target """ def __init__(self, target, host=None): """Construct a TVM target object from 1) Raw target string 2) Target config dict 3) Target tag Parameters ---------- target : Union[str, Dict[str, Any]] Can be one of a literal target string, a json string describing a configuration, or a dictionary of configuration options. When using a dictionary or json string to configure target, the possible values are: kind : str (required) Which codegen path to use, for example 'llvm' or 'cuda'. keys : List of str (optional) A set of strategies that can be dispatched to. When using "kind=opencl" for example, one could set keys to ["mali", "opencl", "gpu"]. device : str (optional) A single key that corresponds to the actual device being run on. This will be effectively appended to the keys. libs : List of str (optional) The set of external libraries to use. For example ['cblas', 'mkl']. system-lib : bool (optional) If True, build a module that contains self registered functions. Useful for environments where dynamic loading like dlopen is banned. mcpu : str (optional) The specific cpu being run on. Serves only as an annotation. model : str (optional) An annotation indicating what model a workload came from. runtime : str (optional) An annotation indicating which runtime to use with a workload. mtriple : str (optional) The llvm triplet describing the target, for example "arm64-linux-android". mattr : List of str (optional) The llvm features to compile with, for example ["+avx512f", "+mmx"]. mfloat-abi : str (optional) An llvm setting that is one of 'hard' or 'soft' indicating whether to use hardware or software floating-point operations. mabi : str (optional) An llvm setting. Generate code for the specified ABI, for example "lp64d". host : Union[str, Dict[str, Any]] (optional) Description for target host. Can be recursive. Similar to target. host : Optional[Union[str, Dict[str, Any]]] Similar to target but for target host. Can be one of a literal target host string, a json string describing a configuration, or a dictionary of configuration options. When using a dictionary or json string to configure target, the possible values are same as target. """ if isinstance(target, str) and "-libs=mkldnn" in target: target = target.replace("mkldnn", "dnnl") warnings.warn( "Legacy support of mkldnn is going to be deprecated. " "Please use -libs=dnnl instead.", ) if isinstance(target, (dict, str)): target = convert(target) if isinstance(host, (dict, str)): host = convert(host) if target is None or not isinstance(target, (Map, String, Target)): raise ValueError("target has to be a string or dictionary.") if host is not None: if not isinstance(host, (Map, String, Target)): raise ValueError("target host has to be a string or dictionary.") self.__init_handle_by_constructor__(_ffi_api.Target, Target(target), Target(host)) else: self.__init_handle_by_constructor__(_ffi_api.Target, target) def __enter__(self): _ffi_api.TargetEnterScope(self) return self def __exit__(self, ptype, value, trace): _ffi_api.TargetExitScope(self) def export(self): return _ffi_api.TargetExport(self) def with_host(self, host=None): return _ffi_api.WithHost(self, Target(host)) @staticmethod def current(allow_none=True): """Returns the current target. Parameters ---------- allow_none : bool Whether allow the current target to be none Raises ------ ValueError if current target is not set. """ return _ffi_api.TargetCurrent(allow_none) @property def arch(self): """Returns the cuda arch from the target if it exists.""" return str(self.attrs.get("arch", "")) @property def max_num_threads(self): """Returns the max_num_threads from the target if it exists.""" return int(self.attrs["max_num_threads"]) @property def max_block_size_x(self): """Returns the max block size in x-dimension from the target if it exists.""" return int(self.attrs["max_block_size_x"]) @property def max_block_size_y(self): """Returns the max block size in y-dimension from the target if it exists.""" return int(self.attrs["max_block_size_y"]) @property def thread_warp_size(self): """Returns the thread_warp_size from the target if it exists.""" return int(self.attrs["thread_warp_size"]) @property def max_shared_memory_per_block(self): return int(self.attrs["max_shared_memory_per_block"]) @property def max_function_args(self): return int(self.attrs.get("max_function_args", -1)) @property def vtcm_capacity(self): return int(self.attrs.get("vtcm-capacity", 0)) @property def device_name(self): return str(self.attrs.get("device", "")) @property def model(self): """Returns model from the target if it exists.""" return str(self.attrs.get("model", "unknown")) @property def mcpu(self): """Returns the mcpu from the target if it exists.""" return str(self.attrs.get("mcpu", "")) @property def mattr(self): """Returns the mattr from the target if it exists.""" return list(self.attrs.get("mattr", [])) @property def supports_integer_dot_product(self): if self.attrs.get("supports_integer_dot_product", []): return bool(self.attrs["supports_integer_dot_product"]) if self.kind.name == "cuda": sm_version = int(self.arch.split("_")[1]) if sm_version >= 61: return True return False @property def libs(self): return list(self.attrs.get("libs", [])) @property def supports_cooperative_matrix(self): if self.attrs.get("supports_cooperative_matrix", []): return bool(self.attrs["supports_cooperative_matrix"]) else: return False @property def features(self): return TargetFeatures(self) @property def l2_cache_size_bytes(self): return int(self.attrs.get("l2_cache_size_bytes", 0)) def get_kind_attr(self, attr_name): """Get additional attribute about the target kind. Parameters ---------- attr_name : str The attribute name. Returns ------- value : object The attribute value """ return _ffi_api.TargetKindGetAttr(self.kind, attr_name) def get_target_device_type(self): """Returns the device_type for this target.""" return _ffi_api.TargetGetDeviceType(self) @staticmethod def list_kinds(): """Returns the list of available target names.""" return list(_ffi_api.ListTargetKinds()) @staticmethod def canon_target(target): """Given a single target-like object, returns the TVM Target object representing it. Can convert from: - None (to None). - An existing TVM Target object. - A string, eg "cuda" or "cuda -arch=sm_80" - A Python dictionary, eg {"kind": "cuda", "arch": "sm_80" } """ if target is None: return None if isinstance(target, Target): return target return Target(target) @staticmethod def canon_target_and_host(target, target_host=None): """Returns a TVM Target capturing target and target_host. Also returns the host in canonical form. The given target can be in any form recognized by Target.canon_target. If given, target_host can be in any form recognized by Target.canon_target. If target_host is given it will be set as the 'host' in the result Target object (and a warning given). Note that this method does not support heterogeneous compilation targets. """ target = Target.canon_target(target) if target is None: assert target_host is None, "Target host is not empty when target is empty." return target, target_host if target.host is None and target_host is not None: warnings.warn( "target_host parameter is going to be deprecated. " "Please pass in tvm.target.Target(target, host=target_host) instead." ) target_host = Target.canon_target(target_host) target = target.with_host(target_host) if target is not None: # In case the target already had a host, extract it here. target_host = target.host return target, target_host @staticmethod def canon_multi_target(multi_targets): """Given a single target-like object, or a collection-like object of target-like objects, returns a TVM Array of TVM Target objects representing then. Can convert from: - None (to None). - A single target-like object in a form recognized by canon_target. - A Python list or TVM Array of target-like objects in a form recognized by canon_target. - A Python dict or TVM Map from TVM IntImm objects representing device types to a target-like object in a form recognized by canon_target. (This is a legacy method to represent heterogeneous targets. The keys are ignored.) """ if multi_targets is None: return None if isinstance(multi_targets, (dict, Map)) and "kind" not in multi_targets: # Convert legacy heterogeneous map representation to ordinary list of targets. return Target.canon_multi_target(list(multi_targets.values())) if isinstance(multi_targets, (list, Array)): # Multiple Target results. return convert([Target.canon_target(tgt) for tgt in multi_targets]) # Single Target result. return convert([Target.canon_target(multi_targets)]) @staticmethod def canon_multi_target_and_host(target, target_host=None): """Returns a TVM Array<Target> capturing target and target_host. The given target can be in any form recognized by Target.canon_multi_target. If given, target_host can be in any form recognized by Target.canon_target. If target_host is given it will be set as the 'host' in each result Target object (and a warning given). """ # Convert target to Array<Target>, but not yet accounting for any host. raw_targets = Target.canon_multi_target(target) assert raw_targets is not None and len(raw_targets) > 0 # Convert host to Target, if given. if raw_targets[0].host is None and target_host is not None: warnings.warn( "target_host parameter is going to be deprecated. " "Please pass in tvm.target.Target(target, host=target_host) instead." ) # Make sure the (canonical) host is captured in all the (canonical) targets. target_host = Target.canon_target(target_host) raw_targets = convert([tgt.with_host(target_host) for tgt in raw_targets]) return raw_targets @staticmethod def canon_target_map_and_host(target_map, target_host=None): """Returns target_map as a map from TVM Target's in canonical form to IRModules. The keys of the input target_map can be in any form recognized by Target.canon_target. Similarly, if given, target_host can be in any form recognized by Target.canon_target. The final target_map keys will capture the target_host in canonical form. Also returns the target_host in canonical form.""" new_target_map = {} canonical_target_host = None for tgt, mod in target_map.items(): tgt = Target.canon_target(tgt) assert tgt is not None if canonical_target_host is None: if tgt.host is not None: canonical_target_host = tgt.host elif target_host is not None: # No deprecation warning in this case since host may have been manufactured # behind the scenes in build_module.py build. canonical_target_host = Target.canon_target(target_host) if tgt.host is None and canonical_target_host is not None: tgt = tgt.with_host(canonical_target_host) new_target_map[tgt] = mod return new_target_map, canonical_target_host @staticmethod def target_or_current(target): """Returns target, or the current target in the environment if target is None""" if target is None: target = Target.current() if target is None: raise ValueError("Target is not set in env or passed as argument.") return target # TODO(@tvm-team): Deprecate the helper functions below. Encourage the usage of config dict instead. def _merge_opts(opts, new_opts): """Helper function to merge options""" if isinstance(new_opts, str): new_opts = new_opts.split() if new_opts: opt_set = set(opts) new_opts = [opt for opt in new_opts if opt not in opt_set] return opts + new_opts return opts def cuda(model="unknown", arch=None, options=None): """Returns a cuda target. Parameters ---------- model: str The model of cuda device (e.g. 1080ti) arch: str The cuda architecture (e.g. sm_61) options : str or list of str Additional options """ opts = _merge_opts(["-model=%s" % model], options) if arch: opts = _merge_opts(["-arch=%s" % arch], opts) if not any(["-arch" in opt for opt in opts]): warnings.warn("Try specifying cuda arch by adding 'arch=sm_xx' to your target.") return Target(" ".join(["cuda"] + opts)) def rocm(model="unknown", options=None): """Returns a ROCM target. Parameters ---------- model: str The model of this device options : str or list of str Additional options """ opts = _merge_opts(["-model=%s" % model], options) return Target(" ".join(["rocm"] + opts)) def mali(model="unknown", options=None): """Returns a ARM Mali GPU target. Parameters ---------- model: str The model of this device options : str or list of str Additional options """ opts = ["-device=mali", "-model=%s" % model] opts = _merge_opts(opts, options) return Target(" ".join(["opencl"] + opts)) def intel_graphics(model="unknown", options=None): """Returns an Intel Graphics target. Parameters ---------- model: str The model of this device options : str or list of str Additional options """ opts = ["-device=intel_graphics", "-model=%s" % model, "-thread_warp_size=16"] opts = _merge_opts(opts, options) return Target(" ".join(["opencl"] + opts)) MICRO_SUPPORTED_MODELS = { "host": [], "atsamd51": ["-mcpu=cortex-m4"], "cxd5602gg": ["-mcpu=cortex-m4"], "esp32": [], "imxrt10xx": ["-mcpu=cortex-m7"], "mps2_an521": ["-mcpu=cortex-m33"], "mps3_an547": ["-mcpu=cortex-m55"], "nrf52840": ["-mcpu=cortex-m4+nodsp"], "nrf5340dk": ["-mcpu=cortex-m33"], "rp2040": ["-mcpu=cortex-m0"], "sam3x8e": ["-mcpu=cortex-m3"], "stm32f746xx": ["-mcpu=cortex-m7", "-march=armv7e-m"], "stm32h7xx": ["-mcpu=cortex-m7"], "stm32l4r5zi": ["-mcpu=cortex-m4"], "stm32u5xx": ["-mcpu=cortex-m33"], "zynq_mp_r5": ["-mcpu=cortex-r5"], } def micro(model="unknown", options=None): """Returns a microTVM target. Parameters ---------- model : str Canonically identifies the target device. This is typically a device board level name. The allowed values are MICRO_SUPPORTED_MODELS.keys(). options : str or list of str Additional options """ if model not in MICRO_SUPPORTED_MODELS: raise ValueError(f"Model {model} not supported by tvm.target.micro.") opts = _merge_opts( MICRO_SUPPORTED_MODELS[model] + [f"-model={model}"], options, ) # NOTE: in the future, the default micro target will be LLVM except when # external dependencies are present. return Target(" ".join(["c"] + opts)) def arm_cpu(model="unknown", options=None): """Returns a ARM CPU target. This function will also download pre-tuned op parameters when there is none. Parameters ---------- model: str SoC name or phone name of the arm board. options : str or list of str Additional options """ trans_table = { "pixel2": ["-model=snapdragon835", "-mtriple=arm64-linux-android", "-mattr=+neon"], "mate10": ["-model=kirin970", "-mtriple=arm64-linux-android", "-mattr=+neon"], "mate10pro": ["-model=kirin970", "-mtriple=arm64-linux-android", "-mattr=+neon"], "p20": ["-model=kirin970", "-mtriple=arm64-linux-android", "-mattr=+neon"], "p20pro": ["-model=kirin970", "-mtriple=arm64-linux-android", "-mattr=+neon"], "rasp3b": ["-model=bcm2837", "-mtriple=armv7l-linux-gnueabihf", "-mattr=+neon"], "rasp4b": [ "-model=bcm2711", "-mtriple=armv8l-linux-gnueabihf", "-mattr=+neon", "-mcpu=cortex-a72", ], "rasp4b64": [ "-model=bcm2711", "-mtriple=aarch64-linux-gnu", "-mattr=+neon", "-mcpu=cortex-a72", ], "rk3399": ["-model=rk3399", "-mtriple=aarch64-linux-gnu", "-mattr=+neon"], "pynq": ["-model=pynq", "-mtriple=armv7a-linux-eabi", "-mattr=+neon"], "ultra96": ["-model=ultra96", "-mtriple=aarch64-linux-gnu", "-mattr=+neon"], "beagleai": [ "-model=beagleai", "-mtriple=armv7a-linux-gnueabihf", "-mattr=+neon,+vfp4,+thumb2", "-mcpu=cortex-a15", ], "stm32mp1": [ "-model=stm32mp1", "-mtriple=armv7a-linux-gnueabihf", "-mattr=+neon,+vfp4,+thumb2", "-mcpu=cortex-a7", ], "thunderx": [ "-model=thunderx", "-mtriple=aarch64-linux-gnu", "-mattr=+neon,+crc,+lse", "-mcpu=thunderxt88", ], } pre_defined_opt = trans_table.get(model, ["-model=%s" % model]) opts = ["-keys=arm_cpu,cpu", "-device=arm_cpu"] + pre_defined_opt opts = _merge_opts(opts, options) return Target(" ".join(["llvm"] + opts)) def rasp(options=None): """Return a Raspberry 3b target. Parameters ---------- options : str or list of str Additional options """ warnings.warn( "tvm.target.rasp() is going to be deprecated. " 'Please use tvm.target.arm_cpu("rasp3b")' ) return arm_cpu("rasp3b", options) def vta(model="unknown", options=None): opts = ["-device=vta", "-keys=vta,cpu", "-model=%s" % model] opts = _merge_opts(opts, options) return Target(" ".join(["ext_dev"] + opts)) def bifrost(model="unknown", options=None): """Return an ARM Mali GPU target (Bifrost architecture). Parameters ---------- options : str or list of str Additional options """ opts = ["-device=bifrost", "-model=%s" % model] opts = _merge_opts(opts, options) return Target(" ".join(["opencl"] + opts)) def riscv_cpu(model="sifive-u54", options=None): """Returns a RISC-V CPU target. Default: sifive-u54 rv64gc Parameters ---------- model: str CPU name. options : str or list of str Additional options """ trans_table = { "sifive-e31": [ "-model=sifive-e31", "-mtriple=riscv32-unknown-linux-gnu", "-mcpu=sifive-e31", "-mabi=ilp32", # cc: riscv64-unknown-linux-gnu-g++ -march=rv32imac -mabi=ilp32 -mcpu=sifive-e31 ], "sifive-e76": [ "-model=sifive-e76", "-mtriple=riscv32-unknown-linux-gnu", "-mcpu=sifive-e76", "-mabi=ilp32", # cc: riscv64-unknown-linux-gnu-g++ -march=rv32imafc -mabi=ilp32 -mcpu=sifive-e76 ], "sifive-u54": [ "-model=sifive-u54", "-mtriple=riscv64-unknown-linux-gnu", "-mcpu=sifive-u54", "-mabi=lp64d", # cc: riscv64-unknown-linux-gnu-g++ -march=rv64gc -mabi=lp64d -mcpu=sifive-u54 ], "sifive-u74": [ "-model=sifive-u74", "-mtriple=riscv64-unknown-linux-gnu", "-mcpu=sifive-u74", "-mabi=lp64d", # cc: riscv64-unknown-linux-gnu-g++ -march=rv64gc -mabi=lp64d -mcpu=sifive-u74 ], } pre_defined_opt = trans_table.get(model, ["-model=%s" % model]) opts = ["-keys=arm_cpu,cpu", "-device=arm_cpu"] + pre_defined_opt opts = _merge_opts(opts, options) return Target(" ".join(["llvm"] + opts)) def hexagon(cpu_ver="v68", **kwargs): """Returns a Hexagon target. Parameters ---------- cpu_ver : str (default: "v68") CPU version used for code generation. Not all allowed cpu str will be valid, LLVM will throw an error. Recognized keyword parameters ----------------------------- hvx : int (default: 128) Size of HVX vector in bytes. Value of 0 disables HVX codegen. llvm_options : str or list of str (default: None) User defined compiler arguments. use_qfloat : bool (default: True for cpu_ver >= v68, False otherwise) Whether to use QFloat HVX instructions. use_ieee_fp : bool (default: False) Whether to use IEEE HVX instructions num_cores : int (default: 4) The number of HVX threads. This attribute is required by meta scheduler. vtcm_capacity: int (default: 0) Hexagon VTCM capacity limitation. If the value is 0, the capacity is treated as unbounded. Note: Floating point support in HVX requires LLVM 14+. """ # Some of the target parameters correspond to target kind attributes # listed in src/target/target_kind.cc. For those parameters, their # names follow the attribute names with the exception of '_' being used # in place of '-'. # Example compiler arguments # llvm -mtriple=hexagon -mcpu=hexagonv68 -mattr=+hvxv68,+hvx-length128b def get_arch_version(cpu_ver): m = re.match(r"v([0-9]+).*", cpu_ver) assert m return int(m.group(1)) # Check for valid codegen cpu valid_hex = ["v65", "v66", "v67", "v67t", "v68", "v69", "v71", "v73"] try: cpu_ver = cpu_ver[cpu_ver.index("v") :].lower() assert cpu_ver in valid_hex except: msg = "{} is not a valid Hexagon version\nvalid versions include {}" raise ValueError(msg.format(cpu_ver, valid_hex)) from None # Target configuration: arch_version = get_arch_version(cpu_ver) config = { "hvx": 128, "llvm_options": None, "use_qfloat": arch_version >= 68, "use_ieee_fp": False, "vtcm_capacity": 0, } config.update(kwargs) # Warn about obsolete parameter names. if config.get("sim_args") or config.get("sim_options"): msg = ( "Setting simulator options in target is deprecated, set environment variable " "HEXAGON_SIM_ARGS instead" ) warnings.warn(msg, stacklevel=2) if config.get("llvm_args"): msg = "The keyword parameter 'llvm_args' is deprecated, use 'llvm_options' instead" warnings.warn(msg, stacklevel=2) config.update({"llvm_options": config["llvm_args"]}) # LLVM target string def create_llvm_target(cpu_ver, config): """Create LLVM target string.""" target = " -mtriple=hexagon" mcpu = " -mcpu=hexagon" + cpu_ver # Process the options that affect target features and return the # target feature string. def create_target_features(config): features = { "use_qfloat": "hvx-qfloat", "use_ieee_fp": "hvx-ieee-fp", } tfs = [] if config["hvx"] > 0: valid_hvx = [0, 64, 128] if not config["hvx"] in valid_hvx: raise ValueError("Invalid hvx value, should be one of " + str(valid_hvx)) tfs += ["+hvx" + cpu_ver, "+hvx-length" + str(config["hvx"]) + "b"] else: tfs += ["-hvx"] # All the additional features happen to only apply to v68+. # Don't bother applying them (even with '-') to lower versions. if arch_version >= 68: tfs += ["-+"[config[f]] + features[f] for f in features] return "-mattr=" + ",".join(tfs) if tfs else "" return target + mcpu + " " + create_target_features(config) # LLVM options string def create_llvm_options(cpu_ver, config): # pylint: disable=unused-argument """Create LLVM options string.""" llvm_options = config["llvm_options"] # To enable auto-vectorization for v68 target added the below llvm-option by default if arch_version == 68: if not llvm_options: llvm_options = "" llvm_options += " -force-hvx-float" # TVM's option parser doesn't allow '=' in values, but '=' can # appear in LLVM flags. Replace it with '@', since it's unlikely # that '@' will be used in another context. if llvm_options is None or len(llvm_options.strip()) == 0: return "" args = [s.replace("=", "@") for s in llvm_options.split()] return "--llvm-options=" + ",".join(args) target_str = create_llvm_target(cpu_ver, config) llvm_str = create_llvm_options(cpu_ver, config) args_list = target_str.split() + llvm_str.split() num_cores = config["num_cores"] if "num_cores" in kwargs else 4 args_list.append("--num-cores=%d" % num_cores) args_list.append("--vtcm-capacity=%d" % config["vtcm_capacity"]) return Target(" ".join(["hexagon"] + args_list)) STM32_SUPPORTED_SERIES = { # High-Performance "stm32H7xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m7", "-march=armv7e-m"], "stm32F7xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m7"], "stm32F4xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m4"], "stm32F2xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m3"], # Mainstream "stm32G0xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m0+"], "stm32F0xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m0"], "stm32F1xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m3"], "stm32G4xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m4"], "stm32F3xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m4"], # Low-power "stm32U5xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m33"], "stm32L5xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m33"], "stm32L4xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m4"], "stm32L1xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m3"], "stm32L0xx": ["-keys=arm_cpu,cpu", "-device=arm_cpu", "-mcpu=cortex-m0+"], } def stm32(series="unknown", options=None): """Returns a STM32 target. Parameters ---------- series: str Series name of a STM32 board series, eg. stm32H7xx or stm32F4xx options : str or list of str Additional options """ if series not in STM32_SUPPORTED_SERIES: raise ValueError(f"Series {series} is not supported by tvm.target.stm32.") opts = _merge_opts(STM32_SUPPORTED_SERIES[series], options) return Target(" ".join(["c"] + opts)) def adreno(model="unknown", options=None): """Returns a Qualcomm GPU target. Parameters ---------- model: str The model of this device options : str or list of str Additional options """ opts = ["-device=adreno", "-model=%s" % model] opts = _merge_opts(opts, options) return Target(" ".join(["opencl"] + opts)) def create(target): """Deprecated. Use the constructor of :py:mod:`tvm.target.Target` directly.""" warnings.warn("tvm.target.create() is being deprecated. Please use tvm.target.Target() instead") return Target(target) @_register_func("target._load_config_dict") def _load_config_dict(config_dict_str): try: config = json.loads(config_dict_str) except json.decoder.JSONDecodeError: return None if not isinstance(config, dict): return None for key in config.keys(): if not isinstance(key, str): return None return config
31,796
36.059441
100
py
tvm
tvm-main/python/tvm/target/intrin.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Target dependent intrinsic registration.""" from tvm.ir import register_intrin_lowering from tvm.tir import call_pure_extern def _rule_float_suffix(op): """Intrinsic rule: Add float suffix if it is float32. This is an example intrinsic generation rule. Parameters ---------- op : PrimExpr The call expression of original intrinsic. Returns ------- ret : PrimExpr The translated intrinsic rule. Return same op if no translation is possible. See Also -------- register_intrin_lowering : The registration function for intrinsic lowering rule. """ name = op.op.name assert name.startswith("tir.") prefix = name[4:] if op.dtype == "float32": return call_pure_extern(op.dtype, "%sf" % prefix, *op.args) if op.dtype == "float64": return call_pure_extern(op.dtype, prefix, *op.args) return op def _rule_float_direct(op): """Intrinsic rule: Directly call pure extern function for floats. This is an example intrinsic generation rule. Parameters ---------- op : PrimExpr The call expression of original intrinsic. Returns ------- ret : PrimExpr The translated intrinsic rule. Return same op if no translation is possible. See Also -------- register_intrin_lowering : The registration function for intrinsic lowering rule. """ if str(op.dtype).startswith("float"): return call_pure_extern(op.dtype, op.op.name[4:], *op.args) return None # opencl pattern for exp register_intrin_lowering("tir.exp", target="opencl", f=_rule_float_direct, level=99) # default pattern for exp register_intrin_lowering("tir.exp", target="default", f=_rule_float_suffix, level=99)
2,551
30.121951
85
py
tvm
tvm-main/python/tvm/target/x86.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Common x86 related utilities""" from .._ffi import register_func from .target import Target @register_func("tvm.target.x86.target_has_sse41") def target_has_sse41(target): return ( target_has_sse42(target) or target_has_avx(target) or target_has_avx2(target) or target_has_avx512(target) or target_has_vnni(target) or target in { "btver2", "penryn", } ) @register_func("tvm.target.x86.target_has_sse42") def target_has_sse42(target): return ( target_has_avx(target) or target_has_avx2(target) or target_has_avx512(target) or target_has_vnni(target) or target in { "silvermont", "slm", "goldmont", "goldmont-plus", "tremont", "nehalem", "corei7", "westmere", "bdver1", "bdver2", "bdver3", "x86-64-v2", } ) @register_func("tvm.target.x86.target_has_avx") def target_has_avx(target): return ( target_has_avx2(target) or target_has_avx512(target) or target_has_vnni(target) or target in {"sandybridge", "corei7-avx", "ivybridge", "core-avx-i"} ) @register_func("tvm.target.x86.target_has_avx2") def target_has_avx2(target): return ( target_has_avx512(target) or target_has_vnni(target) or target in { "haswell", "core-avx2", "broadwell", "skylake", "bdver4", "znver1", "znver2", "znver3", "x86-64-v3", } ) @register_func("tvm.target.x86.target_has_avx512") def target_has_avx512(target): return target in { "skylake-avx512", "skx", "knl", "knm", "x86-64-v4", "cannonlake", # explicit enumeration of VNNI capable due to collision with alderlake "cascadelake", "icelake-client", "icelake-server", "rocketlake", "tigerlake", "cooperlake", "sapphirerapids", } @register_func("tvm.target.x86.target_has_vnni") def target_has_vnni(target): return target in { "cascadelake", "icelake-client", "icelake-server", "rocketlake", "tigerlake", "cooperlake", "sapphirerapids", "alderlake", } @register_func("tvm.target.x86.target_has_amx") def target_has_amx(target): return target in { "sapphirerapids", } @register_func("tvm.topi.x86.utils.get_simd_32bit_lanes") def get_simd_32bit_lanes(): mcpu = Target.current().mcpu fp32_vec_len = 4 if target_has_avx512(mcpu): fp32_vec_len = 16 elif target_has_avx2(mcpu): fp32_vec_len = 8 return fp32_vec_len
3,680
24.741259
78
py
tvm
tvm-main/python/tvm/target/codegen.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Code generation related functions.""" from . import _ffi_api from .target import Target def build_module(mod, target): """Build IRModule into Module. Parameters ---------- mod : tvm.IRModule The ir module. target : str The target module type. Returns ------- module : runtime.Module The corressponding module. """ target = Target(target) if isinstance(target, str) else target return _ffi_api.Build(mod, target) def llvm_lookup_intrinsic_id(name): """Lookup LLVM intrinsic id by name. Parameters ---------- name : str The name of the intrinsic. Returns ------- intrin_id : int The intrinsic id. """ return _ffi_api.llvm_lookup_intrinsic_id(name) def llvm_get_intrinsic_name(intrin_id: int) -> str: """Get the name of an intrinsic for a given id. Parameters ---------- intrin_id : int The id of the intrinsic. Returns ------- name : str The name of the intrinsic. """ return _ffi_api.llvm_get_intrinsic_name(intrin_id) def llvm_version_major(allow_none=False): """Get the major LLVM version. Parameters ---------- allow_none : bool Whether do we allow none. Returns ------- major : int The major LLVM version. """ try: return _ffi_api.llvm_version_major() except AttributeError: if allow_none: return None raise RuntimeError("LLVM version is not available, please check if you built TVM with LLVM")
2,367
24.462366
100
py
tvm
tvm-main/python/tvm/target/virtual_device.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Python bindings for creating VirtualDevices.""" import tvm from tvm.runtime import Object from . import _ffi_api @tvm._ffi.register_object class VirtualDevice(Object): """A compile time representation for where data is to be stored at runtime, and how to compile code to compute it.""" def __init__(self, device=None, target=None, memory_scope="") -> None: if device is None: # The 'unconstrained' device has device type -1 and device id -1. device = tvm.device(-1, -1) self.__init_handle_by_constructor__( _ffi_api.VirtualDevice_ForDeviceTargetAndMemoryScope, device, target, memory_scope ) @property def device_type(self) -> int: return self.device_type_int
1,542
36.634146
94
py
tvm
tvm-main/python/tvm/target/datatype.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Bring Your Own Datatypes custom datatype framework TODO(@gussmith23 @hypercubestart) link to BYODT docs when they exist""" import tvm from tvm.runtime import convert, DataType from tvm.tir.expr import ( Call as _Call, Cast as _Cast, FloatImm as _FloatImm, BinaryOpExpr as _BinaryOpExpr, ) from tvm.tir.op import call_pure_extern from tvm._ffi import register_func as _register_func from tvm.tir import call_intrin def register(type_name, type_code): """Register a custom datatype with the given type name and type code Currently, the type code is manually allocated by the user, and the user must ensure that no two custom types share the same code. Generally, this should be straightforward, as the user will be manually registering all of their custom types. Example: .. code-block:: python # Register a dtype named 'posites2' under type code 130. tvm.target.datatype.register('posites2', 130) Parameters ---------- type_name : str The name of the custom datatype. type_code : int The type's code, which should be >= kCustomBegin. See include/tvm/runtime/data_type.h. """ tvm.runtime._ffi_api._datatype_register(type_name, type_code) def get_type_name(type_code): """Get the type name of a custom datatype from the type code. Note that this only works for custom datatypes registered with tvm.target.datatype.register(). It does not work for TVM-native types. Example: .. code-block:: python tvm.target.datatype.register('posites2', 130) assert tvm.target.datatype.get_type_name(130) == 'posites2' Parameters ---------- type_code : int The type code of the custom datatype. Returns ------- type_name : String The name of the custom datatype. """ return tvm.runtime._ffi_api._datatype_get_type_name(type_code) def get_type_code(type_name): """Get the type code of a custom datatype from its type name Note that this only works for custom datatypes registered with tvm.target.datatype.register(). It does not work for TVM-native types. Example: .. code-block:: python tvm.target.datatype.register('posites2', 130) assert tvm.target.datatype.get_type_code('posites2') == 130 Parameters ---------- type_name : str The type name Returns ------- type_code : int The type code of the custom datatype. """ return tvm.runtime._ffi_api._datatype_get_type_code(type_name) def get_type_registered(type_code): """Returns true if a custom datatype is registered under the given type code Example: .. code-block:: python tvm.target.datatype.register('posites2', 130) assert tvm.target.datatype.get_type_registered(130) Parameters ---------- type_code: int The type code Returns ------- type_registered : bool True if a custom datatype is registered under this type code, and false otherwise. """ return tvm.runtime._ffi_api._datatype_get_type_registered(type_code) def register_op( lower_func, op_name, target, src_type_name, dest_type_name=None, intrinsic_name=None ): """Register a lowering function for a specific operator of a custom datatype At build time, Relay must lower operators over custom datatypes into operators it understands how to compile. For each custom datatype operator which Relay finds while lowering custom datatypes, Relay expects to find a user-defined lowering function. Users register their user-defined lowering functions using this function. Users should use create_lower_func to create their lowering function. It should serve most use-cases. Currently, this will work with Casts, intrinsics (e.g. sqrt, sigmoid), and binary expressions (e.g. Add, Sub, Mul, Div). See the LowerCustomDatatypes pass to see how registered functions are used. Lowering Functions ------------------ TODO(@gussmith23) Get the terminology right here. Lowering functions take in a Relay node, and should return a semantically equivalent Relay node which Relay can build. This means that the returned node should not contain any custom datatypes. Users should likely not need to define lowering functions by hand -- see the helper function create_lower_func. Parameters ---------- lower_func : function The lowering function to call. See create_lower_func. op_name : str The name of the operation which the function computes, given by its class name (e.g. Add, LE, Cast, Call). target : str The name of codegen target. src_type_name : str The name of the custom datatype, e.g. posites2 (but not custom[posites2]32). If op_name is not "Cast", then target type is guaranteed to be the same as src_type_name. dest_type_name : str If op_name is "Cast", then this is required and should be set to the dest datatype of the argument to the Cast. If op_name is not "Cast", this is unused. intrinsic_name : str If op_name is "Call" and intrinsic_name is not None, then we assume the op is a Call to an Intrinsic, and intrinsic_name is the intrinsic's name. """ if op_name == "Cast": assert dest_type_name is not None lower_func_name = ( "tvm.datatype.lower." + target + "." + op_name + "." + dest_type_name + "." + src_type_name ) elif op_name == "Call" and intrinsic_name is not None: lower_func_name = ( "tvm.datatype.lower." + target + "." + op_name + ".intrin." + intrinsic_name + "." + src_type_name ) else: lower_func_name = "tvm.datatype.lower." + target + "." + op_name + "." + src_type_name tvm._ffi.register_func(lower_func_name, lower_func) def register_min_func(func, type_name): """Register the function that returns the minimum representable value of type_name. Operators such as max pooling and argmax require the minimum finite value representable by the datatype the op operating on. Users can use this function to register a function that returns a TIR expression node outputting the minimum representable value of their custom data type. Users should use create_min_lower_func to create their lowering function. It should serve most use-cases. Note: for special cases when it is known that the custom datatype is representable by a float, the user can create their own lowering func that returns a FloatImm. The benefits are allowing optimizations such as rewrites to work as expected on custom datatypes. Parameters ---------- func : function Input is an integer num_bits, should return a TIR expression node that represents a scalar tensor of type custom[type_name]num_bits with the minimum representable value. type_name : str The name of the custom datatype, e.g. posites2 (but not custom[posites2]32). """ _register_func("tvm.datatype.min." + type_name, func) def create_min_lower_func(extern_func_map, type_name): """Returns a lowering function for getting the minimum value of a custom datatype. Parameters ---------- extern_func_map : map A map from bit lengths to the name of the extern "C" function to lower to. type_name : string The name of the custom datatype, e.g. posites2 (but not custom[posites2]32). """ def lower(num_bits): dtype = f"custom[{type_name}]{num_bits}" if num_bits not in extern_func_map: raise RuntimeError("missing minimum function for {dtype}") return call_pure_extern(dtype, extern_func_map[num_bits]) return lower def create_lower_func(extern_func_map): """Returns a function which lowers an operation to a function call. Parameters ---------- extern_func_map : map If lowering a Cast, extern_func_map should be a map from tuples of (src_bit_length, dest_bit_length) to the name of the extern "C" function to lower to. Otherwise, for unary and binary ops, it should simply be a map from bit_length to the name of the extern "C" function to lower to. """ def lower(op): """ Takes an op---either a Cast, Call, or a binary op (e.g. an Add) and returns a call to the specified external function, passing the op's argument or arguments. The return type of the call depends on the type of the op: if it is a custom type, then a uint of the same width as the custom type is returned. Otherwise, the type is unchanged.""" dtype = op.dtype t = DataType(dtype) if get_type_registered(t.type_code): dtype = "uint" + str(t.bits) if t.lanes > 1: dtype += "x" + str(t.lanes) key = t.bits if isinstance(op, _Cast): src_bits = DataType(op.value.dtype).bits key = (src_bits, t.bits) if key not in extern_func_map: raise RuntimeError(f"missing key {key} in extern_func_map for {op.astext()}") if isinstance(op, _Cast): return call_pure_extern(dtype, extern_func_map[key], op.value) if isinstance(op, _FloatImm): return call_pure_extern(dtype, extern_func_map[key], op.value) if isinstance(op, _Call): return call_pure_extern(dtype, extern_func_map[key], *op.args) if isinstance(op, _BinaryOpExpr): return call_pure_extern(dtype, extern_func_map[key], op.a, op.b) raise RuntimeError(f"lowering unsupported op: {op.astext()}") return lower def lower_ite(ite_op): """Lowered if then else function that calls intrinsic if_then_else. Unlike a function lowered by create_lower_func, this function calls the tvm intrinsic if_then_else. Parameters ---------- ite_op : Op Takes an if then else op and returns a call to tir.if_then_else function, passing the op's arguments. The return type of the call if a uint of the same width as the custom type is returned. """ dtype = ite_op.dtype t = tvm.DataType(dtype) assert get_type_registered(t.type_code) dtype = "uint" + str(t.bits) if t.lanes > 1: dtype += "x" + str(t.lanes) return call_intrin( dtype, "tir.if_then_else", convert(ite_op.args[0]), convert(ite_op.args[1]), convert(ite_op.args[2]), ) def lower_call_pure_extern(op): """Lowered call pure extern function that calls intrinsic call_pure_extern. Unlike a function lowered by create_lower_func, this function calls the tvm intrinsic call_pure_extern. Parameters ---------- ite_op : Op Takes a call_pure_extern op and returns a call to tir.call_pure_extern function, passing the op's arguments. The return type of the call if a uint of the same width as the custom type is returned. """ dtype = op.dtype t = tvm.DataType(dtype) assert get_type_registered(t.type_code) dtype = "uint" + str(t.bits) if t.lanes > 1: dtype += "x" + str(t.lanes) return call_intrin(dtype, "tir.call_pure_extern", *op.args)
12,348
32.375676
97
py
tvm
tvm-main/python/tvm/target/tag.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Target tags""" from typing import Any, Dict, Optional from . import _ffi_api from .target import Target def list_tags() -> Optional[Dict[str, Target]]: """Returns a dict of tags, which maps each tag name to its corresponding target. Returns ------- tag_dict : Optional[Dict[str, Target]] The dict of tags mapping each tag name to its corresponding target. None if TVM is built in runtime-only mode. """ if hasattr(_ffi_api, "TargetTagListTags"): return _ffi_api.TargetTagListTags() return None def register_tag(name: str, config: Dict[str, Any], override: bool = False) -> Optional[Target]: """Add a user-defined tag into the target tag registry. Parameters ---------- name: str Name of the target, e.g. "nvidia/gtx1080ti" config : Dict[str, Any] The config dict used to create the target override: bool A boolean flag indicating if overriding existing tags are allowed. If False and the tag has been registered already, an exception will be thrown. Returns ------- target : Optional[Target] The target corresponding to the tag None if TVM is built in runtime-only mode. Examples -------- .. code-block:: python register_tag("nvidia/gtx1080ti", config={ "kind": "cuda", "arch": "sm_61", }) """ if hasattr(_ffi_api, "TargetTagAddTag"): return _ffi_api.TargetTagAddTag(name, config, override) return None # To check the correctness of all registered tags, the call is made in library loading time. list_tags() # We purposely maintain all tags in the C++ side to support pure C++ use cases, # and the Python API is only used for fast prototyping. register_tag( "nvidia/gtx1080ti", config={ "kind": "cuda", "arch": "sm_61", }, )
2,661
31.463415
96
py
tvm
tvm-main/python/tvm/target/_ffi_api.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """FFI APIs for tvm.target""" import tvm._ffi tvm._ffi._init_api("target", __name__)
872
38.681818
62
py
tvm
tvm-main/python/tvm/target/generic_func.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Generic function.""" import tvm._ffi try: from decorator import decorate except ImportError: # Allow decorator to be missing in runtime if not tvm._ffi.base._RUNTIME_ONLY: raise from tvm.runtime import Object from .target import Target from . import _ffi_api @tvm._ffi.register_object class GenericFunc(Object): """GenericFunc node reference. This represents a generic function that may be specialized for different targets. When this object is called, a specialization is chosen based on the current target. Note ---- Do not construct an instance of this object, it should only ever be used as a return value from calling into C++. """ def __call__(self, *args): return _ffi_api.GenericFuncCallFunc(self, *args) def set_default(self, func, allow_override=False): """Set the default function to be used if no specializations match the current target. Parameters ---------- func : function The default function allow_override : bool Whether to allow the current default to be overridden """ _ffi_api.GenericFuncSetDefault(self, func, allow_override) def register(self, func, key_list, allow_override=False): """Register a specialization for this GenericFunc. Parameters ---------- func : function The function to be registered. key : str or list of str The key to be registered. allow_override : bool, optional Whether to allow existing keys to be overridden. """ key_list = [key_list] if isinstance(key_list, str) else key_list _ffi_api.GenericFuncRegisterFunc(self, func, key_list, allow_override) def get_packed_func(self): """Get the packed function specified for the current target. Returns ------- func : PackedFunc The function specified for the current target. Return the default function if no specializations match the current target. """ return _ffi_api.GenericFuncGetPackedFunc(self) def get_native_generic_func(name): """Get a generic function from the global registry. If no function is registered under the given name, a new generic function is created. Parameters ---------- name : string The name of the generic function to get Returns ------- func : GenericFunc The generic function for the given name """ return _ffi_api.GenericFuncGetGlobal(name) def override_native_generic_func(func_name): """Override a generic function defined in C++ Generic function allows registration of further functions that can be dispatched on current target context. If no registered dispatch is matched, the fdefault will be called. Parameters ---------- func_name : string The name of the generic func to be overridden Returns ------- fgeneric : function A wrapped generic function. Example ------- .. code-block:: python import tvm # wrap function as target generic @tvm.target.override_native_generic_func("my_func") def my_func(a): return a + 1 # register specialization of my_func under target cuda @my_func.register("cuda") def my_func_cuda(a): return a + 2 # displays 3, because my_func is called print(my_func(2)) # displays 4, because my_func_cuda is called with tvm.target.cuda(): print(my_func(2)) """ generic_func_node = get_native_generic_func(func_name) def fdecorate(fdefault): """Wrap a target generic function, overriding the previous default that was set for the generic function. Parameters ---------- fdefault : function The default function. Returns ------- fgeneric : function A wrapped generic function. """ generic_func_node.set_default(fdefault, allow_override=True) def register(key, func=None, override=True): """Register function to be the dispatch function. Parameters ---------- key : str or list of str The key to be registered. func : function The function to be registered. override : bool, optional Whether override existing registration. Returns ------- The register function is necessary. """ def _do_reg(myf): generic_func_node.register(myf, key, override) return myf if func: return _do_reg(func) return _do_reg def dispatch_func(func, *args, **kwargs): # pylint: disable=unused-argument """The wrapped dispath function""" if kwargs: raise RuntimeError( "Keyword arguments cannot be used when invoking generic_func %s" % func_name ) return generic_func_node(*args) fresult = decorate(fdefault, dispatch_func) fresult.fdefault = fdefault fresult.register = register fresult.generic_func_node = generic_func_node return fresult return fdecorate def generic_func(fdefault): """Wrap a target generic function. Generic function allows registration of further functions that can be dispatched on current target context. If no registered dispatch is matched, the fdefault will be called. Parameters ---------- fdefault : function The default function. Returns ------- fgeneric : function A wrapped generic function. Example ------- .. code-block:: python import tvm # wrap function as target generic @tvm.target.generic_func def my_func(a): return a + 1 # register specialization of my_func under target cuda @my_func.register("cuda") def my_func_cuda(a): return a + 2 # displays 3, because my_func is called print(my_func(2)) # displays 4, because my_func_cuda is called with tvm.target.cuda(): print(my_func(2)) """ dispatch_dict = {} func_name = fdefault.__name__ def register(key, func=None, override=False): """Register function to be the dispatch function. Parameters ---------- key : str or list of str The key to be registered. func : function The function to be registered. override : bool Whether override existing registration. Returns ------- The register function is necessary. """ def _do_reg(myf): key_list = [key] if isinstance(key, str) else key for k in key_list: if k in dispatch_dict and not override: raise ValueError("Key is already registered for %s" % func_name) dispatch_dict[k] = myf return myf if func: return _do_reg(func) return _do_reg def dispatch_func(func, *args, **kwargs): """The wrapped dispatch function""" target = Target.current() if target is None: return func(*args, **kwargs) for k in target.keys: if k in dispatch_dict: return dispatch_dict[k](*args, **kwargs) return func(*args, **kwargs) def get_packed_func(): """The wrapped to get dispatched function""" target = Target.current() if target is None: return fdefault for k in target.keys: if k in dispatch_dict: return dispatch_dict[k] return fdefault fdecorate = decorate(fdefault, dispatch_func) fdecorate.register = register fdecorate.fdefault = fdefault fdecorate.dispatch_dict = dispatch_dict fdecorate.get_packed_func = get_packed_func return fdecorate
8,938
28.308197
96
py
tvm
tvm-main/python/tvm/target/compilation_config.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Python bindings for creating CompilationConfigs.""" import tvm from . import _ffi_api def make_compilation_config(ctxt, target, target_host=None): """Returns a CompilationConfig appropriate for target and target_host, using the same representation conventions as for the standard build interfaces. Intended only for unit testing.""" raw_targets = tvm.target.Target.canon_multi_target_and_host(target, target_host) return _ffi_api.MakeCompilationConfig(ctxt, raw_targets)
1,281
44.785714
91
py
tvm
tvm-main/python/tvm/target/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """Target description and codgen module. TVM's target string is in format ``<target_kind> [-option=value]...``. Note ---- The list of options include: - **-device=<device name>** The device name. - **-mtriple=<target triple>** Specify the target triple, which is useful for cross compilation. - **-mcpu=<cpuname>** Specify a specific chip in the current architecture to generate code for. By default this is infered from the target triple and autodetected to the current architecture. - **-mattr=a1,+a2,-a3,...** Override or control specific attributes of the target, such as whether SIMD operations are enabled or not. The default set of attributes is set by the current CPU. - **-mabi=<abi>** Generate code for the specified ABI, for example "lp64d". - **-system-lib** Build TVM system library module. System lib is a global module that contains self registered functions in program startup. User can get the module using :any:`tvm.runtime.system_lib`. It is useful in environments where dynamic loading api like dlopen is banned. The system lib will be available as long as the result code is linked by the program. We can use :py:func:`tvm.target.Target` to create a tvm.target.Target from the target string. We can also use other specific function in this module to create specific targets. """ from .target import Target, create, TargetKind from .target import ( cuda, rocm, mali, intel_graphics, arm_cpu, rasp, vta, bifrost, riscv_cpu, hexagon, stm32, ) from .virtual_device import VirtualDevice from .compilation_config import make_compilation_config from .tag import list_tags from .generic_func import GenericFunc from .generic_func import generic_func, get_native_generic_func, override_native_generic_func from . import datatype from . import codegen
2,652
31.353659
93
py
tvm
tvm-main/python/tvm/auto_scheduler/measure_record.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name, pointless-string-statement """ Serialization and other I/O support for measurement records (tuning logs). """ import argparse import logging import os import itertools import numpy as np import tvm._ffi from tvm.runtime import Object from .measure import MeasureErrorNo, MeasureCallback from .utils import calc_workload_dis_factor, decode_workload_key from . import _ffi_api logger = logging.getLogger("auto_scheduler") @tvm._ffi.register_object("auto_scheduler.RecordToFile") class RecordToFile(MeasureCallback): """ A measurement callback that writes measurement records into a file. Parameters ---------- filename : str File name for this callback to write log to. """ def __init__(self, filename): dirname = os.path.dirname(os.path.abspath(filename)) if not os.path.exists(dirname): os.makedirs(dirname) self.__init_handle_by_constructor__(_ffi_api.RecordToFile, filename) @tvm._ffi.register_object("auto_scheduler.RecordReader") class RecordReader(Object): """ Reader of the json log file. Parameters ---------- filename : str File name for this reader to load log from. """ def __init__(self, filename): if not os.path.exists(filename): logger.warning("%s does not exist!", filename) # a set to prevent print duplicated message self.messages = set() self.__init_handle_by_constructor__(_ffi_api.RecordReader, filename) def check_workload_key(self, inputs): """Check and throw warnings for records with old format workload key. Parameters ---------- inputs: List[MeasureInput] The measure inputs to be checked. Notes ----- This checker could be deprecated in the future. """ for inp in inputs: _, args = decode_workload_key(inp.task.workload_key) if args is None: continue if not args: msg = ( "MeasureInput with old format workload key %s should be updated " "using the script from https://github.com/apache/tvm/pull/7317." % inp.task.workload_key ) if msg not in self.messages: self.messages.add(msg) logger.warning(msg) def read_lines(self, max_lines=None, skip_lines=0): """Read multiple lines from the log file. Parameters ---------- max_lines : Optional[int] The maximum number of lines. None to read all lines. skip_lines : int = 0 Skip the first n lines. Returns ------- inputs : List[auto_scheduler.measure.MeasureInput] The MeasureInputs loaded from the log file. results : List[auto_scheduler.measure.MeasureResult] The MeasureResults loaded from the log file. Notes ----- Some unimportant and expensive fields in the returned MeasureInput are not deserialized for faster read speed (e.g. input.task.compute_dag, input.state.stages). If you want to use them, you can call the :code:`recover_measure_input` below to rebuild these fields. """ inputs, results = _ffi_api.RecordReaderReadLines( self, max_lines if max_lines else -1, skip_lines ) self.check_workload_key(inputs) return inputs, results def __iter__(self): while True: ret = _ffi_api.RecordReaderReadNext(self) if not ret: break self.check_workload_key([ret[0]]) yield ret[0], ret[1] # (input, result) def load_record_from_string(record): """ Load the measure record from string. Parameters ---------- record: str A record string, including the serialized MeausreInput and MeasureResult. Returns ------- ret: Tuple[MeasureInput, MeasureResult] A tuple of MeasureInput, MeasureResult. """ return _ffi_api.ReadMeasureRecord(record) def dump_record_to_string(inp, res): """ Dump the measure record to a string. Parameters ---------- inp: MeasureInput The measure input. res: MeasureResult The measure result. Returns ------- ret: str The dumped string. """ return _ffi_api.WriteMeasureRecords(inp, res) def load_records(filename): """ Load measurement records from a file. Parameters ---------- filename : str File name to load log from. Returns ------- logs : List[auto_scheduler.measure.MeasureInput, auto_scheduler.measure.MeasureResult] Notes ----- Some unimportant and expensive fields in the returned MeasureInput are not deserialized for faster read speed (e.g., input.task.compute_dag, input.state.stages). If you want to use them, you can call the :code:`recover_measure_input` below to rebuild these fields. """ return zip(*RecordReader(filename).read_lines()) def save_records(filename, inputs, results): """ Append measure records to file. Parameters ---------- filename : str File name to write log to. inputs: List[MeasureInputs] The MeasureInputs to be written. results: List[MeasureResults] The MeasureResults to be written. """ dirname = os.path.dirname(os.path.abspath(filename)) if not os.path.exists(dirname): os.makedirs(dirname) _ffi_api.SaveRecords(filename, inputs, results) def load_best_record(filename, workload_key=None, target=None, include_compatible=False): """Return the best measurement pair form a log file. This may return none results if there is no legal measure pair with the specified workload_key/target found from the log file. Parameters ---------- filename : str File name to load log from. workload_key : Optional[str] The workload key of the compute declaration. With `None`, this returns the best measure pair of all workloads. target : Optional[tvm.target.Target] The target device. With `None`, this returns the best measure pair of all target devices. include_compatible: bool When set to True, all compatible records in the log file will be considered. Returns ------- input : auto_scheduler.measure.MeasureInput The best State's MeasureInput from this log fine. result : auto_scheduler.measure.MeasureResult The best State's MeasureResult from this log fine. """ log_reader = RecordReader(filename) best_cost = 1e30 best_inp = None best_res = None for inp, res in log_reader: if res.error_no != MeasureErrorNo.NO_ERROR: continue if target and inp.task.target.kind.name != target.kind.name: continue costs = [v.value for v in res.costs] cost = np.mean(costs) if workload_key is not None: dis_f = calc_workload_dis_factor( decode_workload_key(workload_key), decode_workload_key(inp.task.workload_key) ) if dis_f == float("inf"): continue if not include_compatible and dis_f != 1: continue # Since different workloads have different FLOPS, we multiply the factor to # eliminate this difference, which is basically the concept of throughput. cost *= dis_f if cost < best_cost: best_cost = cost best_inp = inp best_res = res return best_inp, best_res def distill_record_file(in_file, out_file): """ Pick the best entries from a record file and store them to another file. This function distills the useful log entries from a large log file. If out_file already exists, the best entries from both in_file and out_file will be saved. Parameters ---------- in_file: str The filename of input out_file: str or file The filename of output """ # pylint: disable=import-outside-toplevel from .dispatcher import ApplyHistoryBest context = load_records(in_file) dirname = os.path.dirname(os.path.abspath(out_file)) if not os.path.exists(dirname): os.makedirs(dirname) if os.path.isfile(out_file): out_context = load_records(out_file) context = itertools.chain(context, out_context) def measure_input_str_key(inp): return _ffi_api.SerializeMeasureInput(inp) # Dict[target key, # Dict[workload hash, # Dict[workload args, (cost, (MeasureInput, MeasureResult))]]] # Full type: Dict[str, Dict[str, Dict[Tuple, Tuple[float, Tuple[Measureinput, MeasureResult]]]]] best_records = {} for inp, res in context: if res.error_no != 0: continue # Keep the best record for each target and workload. costs = [x.value for x in res.costs if isinstance(x, tvm.tir.expr.FloatImm)] cost = np.mean(costs) for k in inp.task.target.keys: entry, _, workload_args = ApplyHistoryBest.get_workload_entry( best_records, k, inp.task.workload_key ) if workload_args not in entry or cost < entry[workload_args][0]: entry[workload_args] = (cost, (inp, res)) # Remove duplications by multiple target keys. out_records = {} for target_entry in best_records.values(): for workload_entry in target_entry.values(): for _, (inp, res) in workload_entry.values(): out_records[measure_input_str_key(inp)] = (inp, res) inputs = [] results = [] for inp, res in out_records.values(): inputs.append(inp) results.append(res) # create a new file and save the best records open(out_file, "w") save_records(out_file, inputs, results) logger.info("Extract %d best records from %s to %s", len(inputs), in_file, out_file) def main(): """The main function for CLI.""" parser = argparse.ArgumentParser() parser.add_argument("--mode", choices=["distill"], default="distill") parser.add_argument("-i", "--input", type=str, help="input file") parser.add_argument("-o", "--output", type=str, default=None, help="output file") args = parser.parse_args() logging.basicConfig() logger.setLevel(logging.INFO) if args.mode == "distill": args.output = args.output or args.input + ".best.json" distill_record_file(args.input, args.output) """ Usage: * Distill the best entries from a large log file e.g. python -m tvm.auto_scheduler.measure_record --mode distill -i input.json """ if __name__ == "__main__": main()
11,650
30.746594
100
py
tvm
tvm-main/python/tvm/auto_scheduler/workload_registry.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name """ Workload registration and serialization. We use a json string to represent a workload (a computation graph). The format of the string is `[func_name, [args...]]`. The dag should be the return value of this `func_name(*args)`. Rationale: The workload is actually a compute dag defined by tvm dsl. But serializing compute dags and matching them efficiently is not easy. Therefore, we use the above string to encode a compute dag. These strings are efficient for serialization/matching and won't be too long. When we need the dag, we decode the string and call the function, which will return the dag. """ import json import logging import pickle import tvm._ffi from tvm.runtime._ffi_node_api import LoadJSON, SaveJSON from .utils import deserialize_args, get_func_name, serialize_args logger = logging.getLogger("auto_scheduler") # Global workload function and hash key registry # It stores two types of workload: # 1. User registered tasks. This type of workload is registered # by the decorator "register_workload" # 2. Extracted tasks from a relay program. This type of workload is # registered by function "register_workload_tensors". # # For 1, the dictionary maps a function name to its function pointer # For 2, the dictionary maps a hash key to a list of input/output tensors WORKLOAD_FUNC_REGISTRY = {} def register_workload(func_name, f=None, override=False): """Register a function that generates a certain workload. The input function should take hashable and jsonable arguments (int, float, tuple of int, tvm.tensor.Tensor, ...) and return a list of tvm.tensor.Tensor. Parameters ---------- func_name : Union[Function, str] The generation function that returns the compute declaration Tensors or its function name. f : Optional[Function] The generation function to be registered. override : boolean = False Whether to override existing entry. Examples -------- .. code-block:: python @auto_scheduler.register_workload def matmul(N, M, K): A = te.placeholder((N, K), name='A') B = te.placeholder((K, M), name='B') k = te.reduce_axis((0, K), name='k') C = te.compute((N, M), lambda i, j: tvm.sum(A[i][k] * B[k][j], axis=[k]), name='C') return [A, B, C] """ global WORKLOAD_FUNC_REGISTRY if callable(func_name): f = func_name func_name = get_func_name(f) if not isinstance(func_name, str): raise ValueError("expect string function name") def register(myf): """internal register function""" if func_name in WORKLOAD_FUNC_REGISTRY and not override: raise RuntimeError(f"{func_name} has been registered already") WORKLOAD_FUNC_REGISTRY[func_name] = myf return myf if f: return register(f) return register def register_workload_tensors(workload_key, tensors, override=True): """Register a workload by provding input/output tensors. Since this function is used when extracting/deserializing tasks, it expects duplicated registrations by default. Parameters ---------- workload_key: str The wokrload key of the compute DAG in JSON string. tensors: List[Tensor] The input/output tensors of a compute DAG override : boolean = True Whether to override existing entry. Returns ------- workload_key: str The wokrload key of the compute DAG in JSON string. """ register_workload(workload_key, override=override)(tensors) return workload_key def make_workload_key(func, args): """Make a workload key by function and arguments. Parameters ---------- func : Union[Function, str] The function that returns the compute declaration Tensors. Can be the a function or the function name. args : Args The args of the function. Returns ------- workload_key : str The workload key of the function. """ global WORKLOAD_FUNC_REGISTRY if callable(func): func_name = get_func_name(func) elif isinstance(func, str): func_name = func else: raise ValueError( "Invalid function: " + str(func) + " . `make_workload_key` expects a callable function or its function name" ) if not func_name in WORKLOAD_FUNC_REGISTRY: raise ValueError( f"{func} is not registered. " f"Please register it with @auto_scheduler.register_workload" ) args = serialize_args(args) return json.dumps((func_name,) + args) @tvm._ffi.register_func("auto_scheduler.workload_key_to_tensors") def workload_key_to_tensors(workload_key): """Get the input/output tensors from the workload key. This method is usually used to create a ComputeDAG by workload key. Parameters ---------- workload_key : str The input workload key in JSON string. The format is either (func_name, arguments...) for compute functions, or (hash, shapes...) for ComputeDAG. Returns ------- tensors : List[Tensor] The registered compute declaration Tensors. """ global WORKLOAD_FUNC_REGISTRY # We register ComputeDAG with both hash and argumetns, which are fixed in ComputeDAG, # so we use an entire workload key to query the ComputeDAG. if workload_key in WORKLOAD_FUNC_REGISTRY: return WORKLOAD_FUNC_REGISTRY[workload_key] # We register compute function with only the function name since # it does not bind to specific arguments, so we use the function name to query # the function and call the function with arguments to get the tensors. workload = json.loads(workload_key) name = workload[0] value = WORKLOAD_FUNC_REGISTRY[name] assert callable(value) args = deserialize_args(workload[1:]) result = value(*args) if isinstance(result, tuple): result = list(result) return result def serialize_workload_registry_entry(workload_key): """ Serialize a workload registry entry. This is used when the start method of multiprocessing is spawn. We need to serialize the entry and register it in the new processes. Parameters ---------- workload_key : str The workload key Returns ------- data: Tuple The serialized pickable data """ global WORKLOAD_FUNC_REGISTRY if workload_key in WORKLOAD_FUNC_REGISTRY: sname = workload_key else: workload = json.loads(workload_key) sname = workload[0] svalue = WORKLOAD_FUNC_REGISTRY[sname] if not callable(svalue): # pylint: disable=assignment-from-no-return svalue = SaveJSON(svalue) return sname, svalue def deserialize_workload_registry_entry(data): """ Deserialize a workload registry entry. This should be used along with :code:`serialize_workload_registry_entry` Parameters ---------- data: Tuple The return value of :code:`serialize_workload_registry_entry` """ global WORKLOAD_FUNC_REGISTRY name, value = data if name not in WORKLOAD_FUNC_REGISTRY: # pylint: disable=assignment-from-no-return if not callable(value): value = LoadJSON(value) WORKLOAD_FUNC_REGISTRY[name] = value def save_workload_func_registry(filename): """Dump workload function registry to a pickle binary file. Parameters ---------- filename : str The filename to dump workload function registry to. """ global WORKLOAD_FUNC_REGISTRY pickle.dump(WORKLOAD_FUNC_REGISTRY, open(filename, "wb")) def load_workload_func_registry(filename): """Load workload function registry from a pickle binary file. Parameters ---------- filename : str The filename to load workload function registry from. """ global WORKLOAD_FUNC_REGISTRY WORKLOAD_FUNC_REGISTRY = pickle.load(open(filename, "rb"))
8,827
30.41637
98
py
tvm
tvm-main/python/tvm/auto_scheduler/relay_integration.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=unused-variable,invalid-name """ Integrate auto_scheduler into relay. It implements the following items: 1. Extract search tasks from a relay program 2. Provide auto-scheduling for all TOPI compute functions """ import json import logging import threading import traceback import tvm from tvm import autotvm, transform from tvm._ffi.base import TVMError from tvm.ir.transform import PassContext from tvm.runtime import convert_to_object from tvm.target import Target from tvm.te.tensor import ComputeOp, PlaceholderOp, Tensor from tvm.tir import Reduce from tvm.tir import expr as _expr from . import _ffi_api from .compute_dag import ComputeDAG, LayoutRewriteOption from .dispatcher import DispatchContext from .search_task import SearchTask from .utils import get_const_tuple from .workload_registry import register_workload_tensors logger = logging.getLogger("auto_scheduler") def call_all_topi_funcs(mod, params, target, error_list, opt_level=3): """Call all TOPI compute to extract auto_scheduler tasks in a Relay program""" # pylint: disable=import-outside-toplevel from tvm import relay # Turn off AutoTVM config not found warnings old_autotvm_silent = autotvm.GLOBAL_SCOPE.silent autotvm.GLOBAL_SCOPE.silent = True with transform.PassContext( opt_level=opt_level, config={ "relay.backend.use_auto_scheduler": True, }, disabled_pass={"AutoSchedulerLayoutRewrite"}, ): compiler = relay.vm.VMCompiler() if params: compiler.set_params(params) mod = tvm.IRModule.from_expr(mod) if isinstance(mod, relay.Function) else mod try: compiler.lower(mod, target) except TVMError: error_list.append(f"{traceback.format_exc()}") finally: autotvm.GLOBAL_SCOPE.silent = old_autotvm_silent def extract_tasks( mod, params, target, target_host=None, hardware_params=None, include_simple_tasks=False, dump_workload_to_dag_log=None, opt_level=3, other_targets=None, ): """Extract tuning tasks from a relay program. Parameters ---------- mod: tvm.IRModule or relay.function.Function The module or function to tune params: dict of str to numpy array The associated parameters of the program target: Union[tvm.target.Target, str] The compilation target target_host: Optional[Union[tvm.target.Target, str]] The host compilation target hardware_params : Optional[HardwareParams] Hardware parameters used for the search tasks include_simple_tasks: bool Whether to extract simple tasks that do not include complicated ops. dump_workload_to_dag_log: Optional[str] A file to dump an association between the workload keys and the actual DAG opt_level : Optional[int] The optimization level of the task extractions. other_targets: Optional[List[tvm.target.Target]] Other targets for call_all_topi_funcs, e.g., cutlass target. Returns ------- tasks: List[SearchTask] The tasks in this network weights: List[int] The weight (i.e. the number of appearance) of extracted tasks """ # pylint: disable=import-outside-toplevel target, target_host = Target.canon_target_and_host(target, target_host) # Run the compiler to collect all TOPI calls during compilation. env = TracingEnvironment( TracingMode.EXTRACT_TASK if include_simple_tasks else TracingMode.EXTRACT_COMPLEX_TASK_ONLY ) dispatch_ctx = DispatchContext.current old_verbose = dispatch_ctx.verbose dispatch_ctx.verbose = 0 targets = [target] if other_targets is not None: targets += other_targets errors = [] with env: # Wrap build call in a new thread to avoid the conflict # between python's multiprocessing and tvm's thread pool build_thread = threading.Thread( target=call_all_topi_funcs, args=(mod, params, targets, errors, opt_level) ) build_thread.start() build_thread.join() if errors: error_strings = ["Task extraction had the following errors:"] + errors raise TVMError("\n".join(error_strings)) dispatch_ctx.verbose = old_verbose # create search tasks tasks = [] weights = [] for wkl_key, (weight, func_names) in env.wkl_key_to_weight.items(): tasks.append( SearchTask( workload_key=wkl_key, target=target, hardware_params=hardware_params, # When auto scheduler is used in end to end network, try to apply layout rewrite # to improve the overall performance layout_rewrite_option=LayoutRewriteOption.get_target_default(target, True), task_inputs=( env.wkl_key_to_input_names[wkl_key] if wkl_key in env.wkl_key_to_input_names else None ), task_inputs_save_to_file=True, desc=",".join(func_names), ) ) weights.append(int(weight)) if dump_workload_to_dag_log is not None: with open(dump_workload_to_dag_log, "w") as f: json.dump({task.workload_key: str(task.compute_dag) for task in tasks}, f) return tasks, weights class TracingMode: """Two modes for tracing""" EXTRACT_TASK = 0 # trace all topi calls to extract tasks # same as EXTRACT_TASK but ignore the task without complex ops EXTRACT_COMPLEX_TASK_ONLY = 1 PREPARE_LAYOUT_REWRITE = 2 # trace topi calls to prepare layout rewrite class TracingEnvironment: """Global environment for tracing all topi function calls""" current = None def __init__(self, tracing_mode): self.tracing_mode = tracing_mode self.relay_disable_build_cache = "false" self.func_name_to_wkl_key = {} self.wkl_key_to_weight = {} self.wkl_key_to_input_names = {} def __enter__(self): TracingEnvironment.current = self return self def __exit__(self, exc_type, exc_val, exc_tb): TracingEnvironment.current = None def add_workload_key(self, func_name, workload_key): """Add the workload key of a search task. Parameters ---------- func_name: str The function name of the task. workload_key: str The workload key of a task. """ self.func_name_to_wkl_key[func_name] = workload_key if workload_key not in self.wkl_key_to_weight: self.wkl_key_to_weight[workload_key] = (0, set()) weight, func_names = self.wkl_key_to_weight[workload_key] func_names.add(func_name) self.wkl_key_to_weight[workload_key] = (weight + 1, func_names) def add_workload_input_names(self, workload_key, input_names): """Add special task inputs to this workload. Parameters ---------- workload_key : str The workload key of a task. input_names : List[str] A list of input names. """ self.wkl_key_to_input_names[workload_key] = input_names @tvm._ffi.register_func("auto_scheduler.enter_layout_rewrite") def enter_layout_rewrite(): """Enter layout rewrite tracing environment""" env = TracingEnvironment(TracingMode.PREPARE_LAYOUT_REWRITE) env.__enter__() @tvm._ffi.register_func("auto_scheduler.exit_layout_rewrite") def exit_layout_rewrite(): """Exit layout rewrite tracing environment""" env = TracingEnvironment.current env.__exit__(None, None, None) def traverse_to_get_io_tensors(outs): """Traverse from a list of output tensors to get input/output tensors and other useful information. Parameters ---------- outs: List[Tensor] The output tensors Returns ------- io_tensors: List[Tensor] The input and output tensors with static shape has_layout_free: bool Whether the compute DAG has layout_free placeholders has_complex_op: bool Whether the topi compute function includes at least one complex (reduce) op """ layout_free_ops = [] inputs = [] has_complex_op = False visited = set() def traverse(t): nonlocal has_complex_op # We cannot directly add tensors to the set, because the comparison of # two tensors with ndim=0 is ambiguous. assert t.handle is not None if t.handle.value in visited: return if isinstance(t.op, PlaceholderOp): inputs.append(t) elif isinstance(t.op, ComputeOp): has_complex_op = has_complex_op or any([isinstance(e, Reduce) for e in t.op.body]) if "layout_free_placeholders" in t.op.attrs: layout_free_ops.append(t.op) for x in t.op.input_tensors: traverse(x) visited.add(t.handle.value) for t in outs: traverse(t) io_tensors = inputs + list(outs) for tensor in io_tensors: # Reject the compute if any of its I/O tensors has dynamic shape. if any([not isinstance(v, int) for v in get_const_tuple(tensor.shape)]): return ([], False, False) return (io_tensors, len(layout_free_ops) > 0, has_complex_op) @tvm._ffi.register_func("auto_scheduler.relay_integration.auto_schedule_topi_compute") def auto_schedule_topi(func_name, outs): """Use auto-scheduler to schedule any topi compute function. Note: This is used internally for relay integration. Do not use this as a general user-facing API. Parameters ---------- func_name: str The name of the function being scheduled. outs: List[Tensor] The output tensors of topi compute functions Returns ------- sch: Optional[te.Schedule] A tuned schedule or none (if not tuned) in the final build mode; None in the tracing mode so that the fallback topi schedule will be used. """ # pylint: disable=import-outside-toplevel from tvm.auto_scheduler.measure import ( # lazily import to avoid recursive dependency prepare_input_map, ) io_tensors, has_layout_free, has_complex_op = traverse_to_get_io_tensors(outs) if not io_tensors: # The compute includes dynamic shapes which are not supported yet. return None try: dag = ComputeDAG(io_tensors) except tvm.error.TVMError as err: logger.info("Failed to create a ComputeDAG for auto_scheduler: %s", str(err)) return None workload_key = dag.workload_key() key = register_workload_tensors(workload_key, io_tensors) target = tvm.target.Target.current() dispatch_ctx = DispatchContext.current state = dispatch_ctx.query(target, key, has_complex_op, dag, func_name) schedule = None env = TracingEnvironment.current if env is None: # in the final build mode if state is None: return None schedule, _ = dag.apply_steps_from_state(state) return schedule if env.tracing_mode in [TracingMode.EXTRACT_TASK, TracingMode.EXTRACT_COMPLEX_TASK_ONLY]: # in the task extraction mode if has_complex_op or env.tracing_mode == TracingMode.EXTRACT_TASK: env.add_workload_key(func_name, key) input_map = prepare_input_map(io_tensors, workload_key) if input_map: env.add_workload_input_names(key, list(input_map.values())) elif env.tracing_mode == TracingMode.PREPARE_LAYOUT_REWRITE: # in prepare_layout_rewrite mode if ( LayoutRewriteOption.get_target_default(target, True) != LayoutRewriteOption.NO_REWRITE and has_layout_free ): if state is None: return None # rewrite the layout and update the context for the new dag new_dag = dag.rewrite_layout_from_state(state) new_key = new_dag.workload_key() if new_key != key: dispatch_ctx.update(target, new_key, state) else: raise ValueError("Invalid tracing mode: " + env.tracing_mode) return schedule @tvm._ffi.register_func("auto_scheduler.relay_integration.te_compiler_update_weights") def te_compiler_update_weights(function_weights): """A callback for updating the weights of extracted tasks. When using the TE compiler that avoids compiling the same function multiple times by caching, all extracted tasks have weight 1, so the TE compiler invokes this callback at the end. In this case, we override existing weights with the use_count in TE compiler cache. Parameters ---------- function_weights: Dict[str, int] Mapping from function names to their weights. """ env = TracingEnvironment.current if env is not None: # Override this map with the weights in the TE compiler. env.wkl_key_to_weight = {} for func_name, weight in function_weights.items(): # If the function name is not in the map, then it means we are not interested in # this function during task extraction (e.g., a function without reduction). if func_name not in env.func_name_to_wkl_key: continue workload_key = env.func_name_to_wkl_key[func_name] if workload_key not in env.wkl_key_to_weight: env.wkl_key_to_weight[workload_key] = (0, set()) # Note that the function appears multiple times in a model will be renamed # to make sure function names are unique, so we use the workload key generated # from the function's TE compute to determine their weights. old_weight, func_names = env.wkl_key_to_weight[workload_key] func_names.add(func_name) env.wkl_key_to_weight[workload_key] = (old_weight + weight, func_names) def tensor_no_check_call(self, *indices): """An indexing function without any check. This is the same as `tvm.te.Tensor::__call__` except that the safety check is removed. """ indices = convert_to_object(indices) args = [] for x in indices: if isinstance(x, _expr.PrimExpr): args.append(x) elif isinstance(x, _expr.IterVar): args.append(x.var) else: raise ValueError("The indices must be expression") return _expr.ProducerLoad(self, args) def remove_index_check(tensor): """Remove the safety check in the indexing function for a tensor. This is done by monkey patching its indexing function. After removing the check, we are allowed to create a temporary wrong IR and fix it later in other places. Parameters ---------- tensor: Tensor The tensor to remove index check. """ # Monkey patch the indexing function tensor.__call__ = tensor_no_check_call.__get__(tensor, Tensor) def rewrite_compute_body(compute_tensor, new_layout): """Rewrite the body of a ComputeOp according to a new layout of a placeholder""" op = compute_tensor.op # Get layout free placeholders layout_free_placeholders = op.attrs["layout_free_placeholders"] assert len(layout_free_placeholders) == 1, "Only support one layout free placeholder" placeholder_op = layout_free_placeholders[0].op # Rewrite the index expression in body body = [] for b in op.body: body.append(_ffi_api.RewriteIndexForNewLayout(placeholder_op, new_layout, b)) op_node = tvm.te._ffi_api.ComputeOp(op.name, op.tag, op.attrs, op.axis, body) num = op_node.num_outputs outputs = tuple(op_node.output(i) for i in range(num)) return outputs[0] if num == 1 else outputs def rewrite_tensor_shape(tensor, shape): """Rewrite the tensor shape""" _ffi_api.RewriteTensorShape(tensor, shape) def is_auto_scheduler_enabled(): """Return whether the auto-scheduler is enabled. Parameters ---------- enabled: bool Whether the auto-scheduler is enabled """ return PassContext.current().config.get( "relay.backend.use_auto_scheduler", False, )
17,027
33.469636
99
py
tvm
tvm-main/python/tvm/auto_scheduler/search_policy.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The search policies of TVM auto-scheduler. The auto-scheduler constructs a search space according to the compute declaration. It then randomly samples programs from the search space and uses evolutionary search with a learned cost model to fine tune the sampled programs. The final optimized programs are sent to actual hardware for measurement. The above process is repeated until the auto-scheduler runs out of time budget. Reference: L. Zheng, C. Jia, M. Sun, Z. Wu, C. Yu, et al. "Ansor : Generating High-Performance Tensor Programs for Deep Learning." (OSDI 2020). """ import random import tvm._ffi from tvm.runtime import Object from .cost_model import RandomModel from . import _ffi_api @tvm._ffi.register_object("auto_scheduler.SearchCallback") class SearchCallback(Object): """Callback function before or after search process""" @tvm._ffi.register_object("auto_scheduler.PreloadMeasuredStates") class PreloadMeasuredStates(SearchCallback): """A SearchCallback to load measured states from the log file for a search policy. This can resume the state of the search policy: - Making sure an already measured state in former searches will never be measured again. - The history states can be used to speed up the search process(e.g. SketchPolicy uses history states as starting point to perform Evolutionary Search). Parameters ---------- filename : str The name of the record file. """ def __init__(self, filename): self.__init_handle_by_constructor__(_ffi_api.PreloadMeasuredStates, filename) @tvm._ffi.register_object("auto_scheduler.PreloadCustomSketchRule") class PreloadCustomSketchRule(SearchCallback): """ A SearchCallback for SketchSearchPolicy that allows users to add custom sketch rule. Notes ----- This is an advanced feature. Make sure you're clear how it works and this should only be used in SketchSearchPolicy. Parameters ---------- meet_condition_func: Callable A function with `(policy, state, stage_id) -> int`. Should return one of the result enumeration. apply_func: Callable A function with `(policy, state, stage_id) -> [[State, int], ...]`. rule_name: str = "CustomSketchRule" The name of this custom sketch rule. """ # Result enumeration of the condition function. PASS = 0 # Skip this rule and continue to try the next rules. APPLY = 1 # Apply this rule and continue to try the next rules. APPLY_AND_SKIP_REST = 2 # Apply this rule and skip the rest rules. def __init__(self, meet_condition_func, apply_func, rule_name="CustomSketchRule"): self.__init_handle_by_constructor__( _ffi_api.PreloadCustomSketchRule, meet_condition_func, apply_func, rule_name ) @tvm._ffi.register_object("auto_scheduler.SearchPolicy") class SearchPolicy(Object): """The base class of search policies.""" def continue_search_one_round(self, num_measure, measurer): """ Continue the search by doing an additional search round. Parameters ---------- num_measure: int The number of programs to measure in this round measurer: ProgramMeasurer The program measurer to measure programs Returns ------- inputs: List[MeasureInput] The inputs of measurments in this search round results: List[MeasureResult] The results of measurments in this search round """ return _ffi_api.SearchPolicyContinueSearchOneRound(self, num_measure, measurer) def set_verbose(self, verbose): """ Set the verbosity level of the search policy. Parameters ---------- verbose: int The verbosity level """ return _ffi_api.SearchPolicySetVerbose(self, verbose) @tvm._ffi.register_object("auto_scheduler.EmptyPolicy") class EmptyPolicy(SearchPolicy): """A simple example of the search policy which always returns the initial naive schedule (state). Parameters ---------- task : SearchTask The SearchTask for the computation declaration. init_search_callbacks : Optional[List[SearchCallback]] Callback functions called before the search process. """ def __init__(self, task, init_search_callbacks=None): self.__init_handle_by_constructor__(_ffi_api.EmptyPolicy, task, init_search_callbacks) @tvm._ffi.register_object("auto_scheduler.SketchPolicy") class SketchPolicy(SearchPolicy): """The search policy that searches in a hierarchical search space defined by sketches. The policy randomly samples programs from the space defined by sketches and use evolutionary search to fine-tune them. Parameters ---------- task : SearchTask The SearchTask for the computation declaration. program_cost_model : CostModel = RandomModel() The cost model to estimate the complete schedules. params : Optional[Dict[str, Any]] Parameters of the search policy. See `src/auto_scheduler/search_policy/sketch_search_policy.h` for the definitions. See `DEFAULT_PARAMS` below to find the default values. seed : Optional[int] Random seed. verbose : int = 1 Verbosity level. 0 for silent, 1 to output information during schedule search. init_search_callbacks : Optional[List[SearchCallback]] Callback functions called before the search process, usually used to do extra initializations. Possible callbacks: - auto_scheduler.PreloadMeasuredStates - auto_scheduler.PreloadCustomSketchRule """ DEFAULT_PARAMS = { "eps_greedy": 0.05, "retry_search_one_round_on_empty": 1, "sample_init_min_population": 50, "sample_init_use_measured_ratio": 0.2, "evolutionary_search_population": 2048, "evolutionary_search_num_iters": 4, "evolutionary_search_mutation_prob": 0.85, "cpu_multi_level_tiling_structure": "SSRSRS", "gpu_multi_level_tiling_structure": "SSSRRSRS", # Notice: the default thread bind policy of GPU assumes the tiling structure to have at # least 3 spatial tiling levels in outermost "max_innermost_split_factor": 64, "max_vectorize_size": 16, "disable_change_compute_location": 0, } def __init__( self, task, program_cost_model=RandomModel(), params=None, seed=None, verbose=1, init_search_callbacks=None, ): if params is None: params = SketchPolicy.DEFAULT_PARAMS else: for key, value in SketchPolicy.DEFAULT_PARAMS.items(): if key not in params: params[key] = value self.__init_handle_by_constructor__( _ffi_api.SketchPolicy, task, program_cost_model, params, seed or random.randint(1, 1 << 30), verbose, init_search_callbacks, ) def generate_sketches(self, print_for_debug=False): """Generate the sketches. This python interface is mainly used for debugging and testing. The actual search is all done in c++. Parameters ---------- print_for_debug : bool = False Whether print out the sketches for debug. Returns ------- sketches : List[State] The generated sketches of this search task. """ sketches = _ffi_api.SketchPolicyGenerateSketches(self) if print_for_debug: for i, s in enumerate(sketches): print("=" * 20 + f" {i} " + "=" * 20) print(s) return sketches def sample_initial_population(self): """Sample initial population. This python interface is mainly used for debugging and testing. The actual search is all done in c++. Returns ------- states: List[State] The sampled states """ states = _ffi_api.SketchPolicySampleInitialPopulation(self) return states def evolutionary_search(self, init_populations, out_size): """Perform evolutionary search. This python interface is mainly used for debugging and testing. The actual search is all done in c++. Parameters ---------- init_populations: List[State] The initial population states out_size : int The size of generated states Returns ------- states: List[State] The generated states """ states = _ffi_api.SketchPolicyEvolutionarySearch(self, init_populations, out_size) return states
9,631
33.898551
97
py
tvm
tvm-main/python/tvm/auto_scheduler/compute_dag.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name """ The auto-scheduler's computational graph and related program analyses. """ import hashlib import json import tvm._ffi from tvm.runtime import Object from tvm.runtime._ffi_node_api import LoadJSON, SaveJSON from . import _ffi_api from .loop_state import State, StateObject from .utils import get_const_tuple from .workload_registry import workload_key_to_tensors class LayoutRewriteOption: """ Options for applying layout rewrite. The NO_REWRITE and INSERT_TRANSFORM_STAGE are expected to be used when tuning a standalone op, and the REWRITE_FOR_PRE_TRANSFORMED is expected to be used when tuning ops inside a network. """ # Do not perform layout rewrite NO_REWRITE = 0 # Insert layout transformation stages for input placeholders in the compute DAG INSERT_TRANSFORM_STAGE = 1 # Do not insert layout transformation stages and assume the input placeholders # are pre-transformed. # Note: The lowered function with this option does not accept the origial input shapes, # so this option must be used along with `AutoSchedulerLayoutRewrite` pass in Relay. REWRITE_FOR_PRE_TRANSFORMED = 2 @staticmethod def get_target_default(target, in_relay_integration=False): """Get the default layout rewrite option for the specified target. Currently we only enable layout rewrite for cpu / mali backend for now Parameters ---------- target: tvm.target.Target The compilation target. in_relay_integration: bool If this check is ask for relay integration. Returns ------- layout_rewrite_option: LayoutRewriteOption The default layout rewrite option for the specified target. """ layout_rewrite_option = LayoutRewriteOption.NO_REWRITE if target.kind.name == "llvm" or ( "device" in target.attrs and target.attrs["device"] == "mali" ): layout_rewrite_option = ( LayoutRewriteOption.REWRITE_FOR_PRE_TRANSFORMED if in_relay_integration else LayoutRewriteOption.INSERT_TRANSFORM_STAGE ) return layout_rewrite_option @tvm._ffi.register_object("auto_scheduler.ComputeDAG") class ComputeDAG(Object): """ The auto-scheduler's computational graph and related program analyses. We convert a compute declaration described by `tvm.compute` (could be a single operator or a subgraph) to a ComputeDAG. It keeps the input/output tensors, all operations in the DAG, and some static analysis results for the DAG (e.g. the total float operation count, consumer/producer relations of operations, whether an operation stage should be tiled/compute inlined). These analyses can help the search policy to make decisions during the search. ComputeDAG is also responsible for the interaction between auto-scheduler's `LoopState` and TVM schedule (e.g. applying the `LoopState` transform steps to a TVM schedule, providing `LoopState` with extra information got from TVM schedule). Parameters ---------- compute : Union[List[Tensor], str, tvm.te.Schedule] Input/output tensors or workload key for a compute declaration. """ def __init__(self, compute_or_sche): if isinstance(compute_or_sche, str): compute = workload_key_to_tensors(compute_or_sche) sche = None elif isinstance(compute_or_sche, (list, tvm.ir.container.Array)): for item in compute_or_sche: if not isinstance(item, tvm.te.Tensor): raise ValueError( "The input of ComputeDAG should be a list of Tensor, but got %s" % type(item) ) compute = compute_or_sche sche = None elif isinstance(compute_or_sche, tvm.te.Schedule): compute = None sche = compute_or_sche else: raise ValueError( "Invalid compute type: %s. ComputeDAG expects string, list of Tensor, or Schedule" % type(compute_or_sche) ) self.__init_handle_by_constructor__(_ffi_api.ComputeDAG, compute, sche) def get_init_state(self): """Get the init state of this ComputeDAG. Returns ------- state : State The initial State without any transform steps. """ return State(self.init_state, self) def apply_steps_from_state(self, state, layout_rewrite=LayoutRewriteOption.NO_REWRITE): """ Apply the history transform steps from a State to get a TVM schedule. Parameters ---------- state : Union[State, StateObject] The state from which we get transform steps. layout_rewrite: LayoutRewriteOption = NoRewrite Rewrite the layout of placeholders specified by "layout_free_placeholders" attr to make it most friendly for the generated schedule to read from. Returns ------- A `te.schedule` and the a list of `te.Tensor` to be used in `tvm.lower` or `tvm.build`. """ state_obj = state if isinstance(state, StateObject) else state.state_object return _ffi_api.ComputeDAGApplyStepsFromState(self, state_obj, layout_rewrite) def print_python_code_from_state(self, state): """ Print transform steps in the history of a State as TVM's python schedule code. This is used to print transformation steps for debugging. Use `apply_steps_from_state` if you want to get a schedule for code generation. Parameters ---------- state : Union[State, StateObject] The state from which we get transform steps. Returns ------- str : Str The Python schedule code. """ state_obj = state if isinstance(state, StateObject) else state.state_object return _ffi_api.ComputeDAGPrintPythonCodeFromState(self, state_obj) def infer_bound_from_state(self, state): """ Infer and fill the bound of all iterators of a state. The states may lose complete bound information after some transform steps (e.g., compute_at). We can call this function to infer and fill all the bound information. This function calls TVM InferBound pass internally to get the bound. The returned state of this function is guaranteed to have complete iterator extent information. Parameters ---------- state : Union[State, StateObject] The state from which we get transform steps. Returns ------- updated_state : State The State with complete bound information. """ state_obj = state if isinstance(state, StateObject) else state.state_object updated_state = State(_ffi_api.ComputeDAGInferBoundFromState(self, state_obj), self) # Copy the stage_id_map from the original state to make sure the old indices are still # valid if isinstance(state, State): for k, v in state.stage_id_map.items(): updated_state.stage_id_map[k] = v return updated_state def rewrite_layout_from_state(self, state): """ Rewrite the layout of the DAG according to the history transform steps of a state. Parameters ---------- state : Union[State, StateObject] The state from which we get transform steps. Returns ------- updated_dag : ComputeDAG The compute dag with rewritten layout. """ state_obj = state if isinstance(state, StateObject) else state.state_object return _ffi_api.ComputeDAGRewriteLayoutFromState(self, state_obj) def workload_key(self): """Return the workload key of this compute DAG. The workload key is a JSON string from a tuple of (hash of DAG, tensor shapes...) Returns ------- key: str The workload key of this compute DAG """ str_dag = _ffi_api.ComputeDAGPrintDAG(self, True) hash_func = tvm._ffi.get_global_func( "auto_scheduler.compute_dag.hash_func", allow_missing=True ) if hash_func is None: str_dag = str_dag.encode("utf-8") hash_key = hashlib.md5(str_dag).hexdigest() else: hash_key = hash_func(str_dag) io_shapes = [] for tensor in self.tensors: io_shapes.append(get_const_tuple(tensor.shape)) return json.dumps([hash_key] + io_shapes) def __str__(self): # pretty print MAX_LINE_WIDTH = 256 raw_lines = super().__str__().split("\n") lines = [] for line in raw_lines: if len(line) > MAX_LINE_WIDTH: line = ( line[: MAX_LINE_WIDTH // 2] + " ..(OMITTED).. " + line[-MAX_LINE_WIDTH // 2 :] ) lines.append(line) return "\n".join(lines) def __getstate__(self): return {"tensors": SaveJSON(self.tensors)} def __setstate__(self, state): # Since we always use tensors to recover the ComputeDAG, we do not support # (de)serialization of the ComputeDAG constructed by a schedule. self.__init_handle_by_constructor__(_ffi_api.ComputeDAG, LoadJSON(state["tensors"]), None) def get_shape_from_rewritten_layout(rewritten_layout, axis_names): """Get the orginal shape from a rewritten layout string. Parameters ---------- rewritten_layout: str The layout after rewrite axis_names: List[str] Specify the order of axes by names Returns ------- shape: List[PrimExpr] The original shape """ return _ffi_api.GetShapeFromRewrittenLayout(rewritten_layout, axis_names)
10,770
36.529617
99
py
tvm
tvm-main/python/tvm/auto_scheduler/task_scheduler.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name """ The task scheduler that allocates the time resources when tuning multiple tasks together The details of the "gradient" strategy below can be found in the section 6 of this paper: L. Zheng, C. Jia, M. Sun, Z. Wu, C. Yu, et al. "Ansor : Generating High-Performance Tensor Programs for Deep Learning." (OSDI 2020). """ import os import time import math import logging import numpy as np from .search_policy import SearchPolicy, SketchPolicy, PreloadMeasuredStates from .cost_model import RandomModel, XGBModel from .utils import array_mean from .measure import ProgramMeasurer from .measure_record import RecordReader from . import _ffi_api logger = logging.getLogger("auto_scheduler") def make_search_policies( search_policy, search_policy_params, tasks, num_measures_per_round, verbose, load_model_file=None, load_log_file=None, adaptive_training=False, ): """Make a list of search policies for a list of search tasks. It creates one policy per task. Parameters ---------- search_policy: Union[str, List[SearchPolicy]] The name of search policy. search_policy_params: Dict[str, Any]] The parameters of the search policy. tasks: List[SearchTask] The list of all tasks num_measures_per_round: int The number of schedules to be measured at each search round. This should be the same as `TuningOptions.num_measures_per_round` verbose: int The verbosity level. 0 for silent. load_model_file: Optional[str] Load pre-trained model from this file. If this is None, the cost model will be trained from scratch. load_log_file: Optional[str] Load measurement records from this file. If it is not None, the status of the task scheduler, search policies and cost models will be restored according to this file. adaptive_training: bool = False Option used by XGBModel to reduce the model training frequency when there're too many logs. Returns ------- policies: List[SearchPolicy] The list of search policies """ if search_policy == "default": search_policy = "sketch.xgb" if isinstance(search_policy, str): policy_type, model_type = search_policy.split(".") if model_type == "xgb": cost_model = XGBModel( num_warmup_sample=len(tasks) * num_measures_per_round, model_file=load_model_file, adaptive_training=adaptive_training, ) if load_model_file and os.path.isfile(load_model_file): logger.info("TaskScheduler: Load pretrained model...") cost_model.load(load_model_file) elif load_log_file: logger.info("TaskScheduler: Reload measured states and train the model...") cost_model.update_from_file(load_log_file) elif model_type == "random": cost_model = RandomModel() else: raise ValueError("Invalid search policy: " + search_policy) if policy_type == "sketch": if load_log_file: # use the log file to restore the status of search policies. init_search_callbacks = [PreloadMeasuredStates(load_log_file)] else: init_search_callbacks = None search_policies = [ SketchPolicy( task, cost_model, params=search_policy_params, verbose=verbose, init_search_callbacks=init_search_callbacks, ) for task in tasks ] else: raise ValueError("Invalid search policy: " + search_policy) else: # check type assert isinstance(search_policy, (tuple, list)) for item in search_policy: assert isinstance(item, SearchPolicy) search_policies = search_policy return search_policies def derive_similarity_tag(dag, log_base=1.618): """Derive the tag for similarity check from one computational DAG. The DAGs with the same tag are considered as similar tasks. The tag format is <op1-tag>_<op2-tag> ... <log(flop)>. If the tag is "", then the task is not considered to be similar to any other tasks. Parameters ---------- dag: ComputeDAG The input computational DAG log_base: float = 1.618 The base of log to normalize FLOPS Returns ------- tag: str The tag of this computational DAG. """ ret = "" for op in dag.ops: tag = op.attrs.get("auto_scheduler_task_scheduler_tag", None) if tag: ret += op.attrs["auto_scheduler_task_scheduler_tag"] + "_" if ret: ret += "%d" % int(math.log(dag.flop_ct + 1, log_base)) return ret class TaskScheduler: """ Allocate the time resources when tuning multiple tasks together. This implements two strategies: "round-robin" and "gradient". Parameters ---------- tasks: List[SearchTask] All tasks to tune task_weights: Optional[List[float]] The weights of tasks. If provided, the task scheduler will set the objective function to sum(weight[t] * latency[t]), where weight[t] is the weight of a task and the lantecy[t] is the lantecy of the task. If not provided, the task scheduer will assign equal weights to all tasks (i.e., the objective function is sum(latency[t])). objective_func: Optional[Callable[List[float] -> float]] The objective function to be minimized. The objective function accepts the current latencies of all tasks and returns the objective. If not provided, the objective is the weighted sum of the latencies of all tasks. strategy: str = "gradient" The scheduling strategy. "round-robin": Tune tasks in round robin order. "gradient" : Tune tasks with gradient descent. load_model_file: Optional[str] Load pre-trained model from this file. If this is None, the cost model will be trained from scratch. load_log_file: Optional[str] Load measurement records from this file. If it is not None, the status of the task scheduler, search policies and cost models will be restored according to this file. verbose: int = 1 The level of verbosity. 0 means silent. alpha: float = 0.2 The parameter used for 'gradient' strategy beta: float = 2 The parameter used for 'gradient' strategy backward_window_size: int = 3 The parameter used for 'gradient' strategy callbacks: Optional[List[TaskSchedulerCallback]] The task scheduler callbacks that will be called before and after tuning a task. If None, PrintTableInfo and LogEstimatedLatency callback will be used. """ def __init__( self, tasks, task_weights=None, objective_func=None, strategy="gradient", load_model_file: str = None, load_log_file: str = None, alpha: float = 0.2, beta: float = 2, gamma: float = 0.5, backward_window_size: int = 3, callbacks=None, ): self.tasks = tasks if objective_func: # use custom objective function self.objective_func = objective_func else: # use weighted sum if task_weights: self.objective_func = lambda costs: sum(c * w for c, w in zip(costs, task_weights)) else: self.objective_func = sum self.strategy = strategy self.load_log_file = load_log_file self.load_model_file = load_model_file self.alpha = alpha self.beta = beta self.gamma = gamma self.backward_window_size = backward_window_size self.callbacks = ( callbacks if callbacks is not None else [PrintTableInfo(), LogEstimatedLatency("total_latency.tsv")] ) assert len(self.tasks) != 0, "No tasks" assert self.strategy in ["round-robin", "gradient"] # task_cts[i] saves how many times task i is tuned self.task_cts = [0 for _ in range(len(self.tasks))] # task_best_cts[i] saves the round task i found the best latency self.task_best_cts = [0 for _ in range(len(self.tasks))] # task_costs_history[i] saves the latency history of task i self.task_costs_history = [[] for _ in range(len(self.tasks))] # best_costs[i] saves the best latency of task i self.best_costs = 1e10 * np.ones(len(self.tasks)) self.cur_score = self._compute_score(self.best_costs) self.tune_option = self.measurer = self.search_policies = None self.ct = self.best_ct = self.best_score = self.tic = None self.num_measures_per_round = None self.dead_tasks = set() # Build similarity groups self.task_tags = [] # task_id -> tag self.tag_to_group_id = {} # tag -> group_id self.group_task_ids = [] # group_id -> all task ids in this group self.flop_cts = [] # task_id -> the number of floating ops for i, task in enumerate(self.tasks): tag = derive_similarity_tag(task.compute_dag) self.task_tags.append(tag) self.flop_cts.append(task.compute_dag.flop_ct) if not tag: continue if tag not in self.tag_to_group_id: self.tag_to_group_id[tag] = len(self.tag_to_group_id) self.group_task_ids.append([]) self.group_task_ids[self.tag_to_group_id[tag]].append(i) def tune( self, tune_option, search_policy="default", search_policy_params=None, adaptive_training=False, per_task_early_stopping=None, ): """Tune a batch of tasks together. Parameters ---------- tune_option: TuningOptions The tuning options applied to all tasks. search_policy: : Union[str, List[SearchPolicy]] = "default" The list of search policies. If it is str, "default" for the default policy (SketchPolicy + XGBModel), "sketch.xgb" for SketchPolicy + XGBModel, "sketch.random" for SketchPolicy + RandomModel. search_policy_params : Optional[Dict[str, Any]] The parameters of the search policy adaptive_training : bool = False Option used by XGBModel to reduce the model training frequency when there're too many logs. per_task_early_stopping : Optional[int] Stop tuning a task early if getting no improvement after n measurements. """ # init members self.tune_option = tune_option self.early_stopping_all = ( 1e20 if tune_option.early_stopping < 0 else tune_option.early_stopping ) self.early_stopping_task = ( 1e20 if per_task_early_stopping is None else per_task_early_stopping ) self.measurer = ProgramMeasurer( tune_option.builder, tune_option.runner, tune_option.measure_callbacks, tune_option.verbose, ) self.ct = self.best_ct = 0 self.tic = time.time() # reset num_measures_per_round to make sure every task is tuned at least once self.num_measures_per_round = min( tune_option.num_measures_per_round, tune_option.num_measure_trials // len(self.tasks) ) if self.num_measures_per_round <= 0: raise ValueError( "num_measure_trials is too small. Please set it to a higher value." f"It should be at least {len(self.tasks)} for this model." ) # restore the status of the task scheduler from a log file if self.load_log_file: self._restore_status(self.load_log_file, self.num_measures_per_round) # make one search policy for one task self.search_policies = make_search_policies( search_policy, search_policy_params, self.tasks, self.num_measures_per_round, tune_option.verbose, self.load_model_file, self.load_log_file, adaptive_training, ) # do a round robin first to warm up for idx in range(len(self.tasks)): # skip warming up this task if it has been tuned before (restored from the log file) if not self.task_cts[idx]: self._tune_task(idx) self.best_ct = self.ct self.best_score = self.cur_score # use the specific strategy to choose workload to tune task_idx = -1 while self.ct < tune_option.num_measure_trials and len(self.dead_tasks) < len(self.tasks): if self.strategy == "round-robin": task_idx = (task_idx + 1) % len(self.tasks) while task_idx in self.dead_tasks: task_idx = (task_idx + 1) % len(self.tasks) elif self.strategy == "gradient": gradients = [] for i in range(len(self.tasks)): if i in self.dead_tasks: gradients.append(0) continue # compute gradient from chain rule : (delta f / delta g_i) delta = 1e-4 new_costs = list(self.best_costs) new_costs[i] -= delta chain_grad = ( self._compute_score(self.best_costs) - self._compute_score(new_costs) ) / delta # compute (g_i(t_i) - g(t_i - \Delta t)) / (\Delta t) if ( self.task_cts[i] - 1 < len(self.task_costs_history[i]) and self.task_cts[i] - 1 - self.backward_window_size >= 0 ): backward_grad = ( self.task_costs_history[i][self.task_cts[i] - 1] - self.task_costs_history[i][ self.task_cts[i] - 1 - self.backward_window_size ] ) / self.backward_window_size else: backward_grad = 0 # compute (g_i(t_i + \Delta t) - g(t_i)) / (\Delta t) g_next_1 = self.best_costs[i] - (self.best_costs[i] / self.task_cts[i]) g_next_2 = self.beta * 1e30 group_id = self.tag_to_group_id.get(self.task_tags[i], None) if group_id is not None and len(self.group_task_ids[group_id]) > 1: best_flops = max( [ self.flop_cts[j] / self.best_costs[j] for j in self.group_task_ids[group_id] ] ) g_next_2 = self.beta * self.flop_cts[i] / best_flops g_next = min(g_next_1, g_next_2) forward_grad = g_next - self.best_costs[i] # combine all grads grad = chain_grad * ( self.alpha * backward_grad + (1 - self.alpha) * forward_grad ) assert grad <= 0 gradients.append(grad) if max(gradients) == min(gradients): task_idx = np.random.choice(len(gradients)) else: task_idx = np.argmin(gradients) else: raise ValueError("Invalid strategy: " + self.strategy) self._tune_task(task_idx) self._adjust_similarity_group(task_idx) if self.cur_score < self.best_score: self.best_score = self.cur_score self.best_ct = self.ct elif self.ct - self.best_ct >= self.early_stopping_all and all( cost < 1e9 for cost in self.best_costs ): if self.tune_option.verbose >= 1: print( "Stop early since no performance improvement in the last " + str(self.early_stopping_all) + " measurement trials." ) break def _tune_task(self, task_idx): """Tune the select task for one round""" # Run pre-tune callbacks for callback in self.callbacks: callback.pre_tune(self, task_idx) measure_inputs, measure_results = self.search_policies[task_idx].continue_search_one_round( self.num_measures_per_round, self.measurer ) self.task_cts[task_idx] += 1 for res in measure_results: cost = array_mean(res.costs) if cost < self.best_costs[task_idx]: self.task_best_cts[task_idx] = self.task_cts[task_idx] self.best_costs[task_idx] = cost # Stop tuning this task in the rest of the process if its search space has been # fully explored or it has no improvement for a long while. no_change_trials = ( self.task_cts[task_idx] - self.task_best_cts[task_idx] ) * self.num_measures_per_round if len(measure_inputs) == 0 or no_change_trials > self.early_stopping_task: self.dead_tasks.add(task_idx) self.task_costs_history[task_idx].append(self.best_costs[task_idx]) self.ct += len(measure_inputs) self.cur_score = self._compute_score(self.best_costs) # Run post-tune callbacks for callback in self.callbacks: callback.post_tune(self, task_idx) def _compute_score(self, costs): """compute the objective function""" # Make sure to return float. score = self.objective_func(costs) return score.value if hasattr(score, "value") else score def _adjust_similarity_group(self, task_idx): """adjust the similarity group for the selected task""" group_id = self.tag_to_group_id.get(self.task_tags[task_idx], None) if group_id is None or len(self.group_task_ids[group_id]) <= 1: return group_ids = self.group_task_ids[group_id] best_group_flops = max([self.flop_cts[j] / self.best_costs[j] for j in group_ids]) cur_flops = self.flop_cts[task_idx] / self.best_costs[task_idx] # if we tune a task for many times but it still cannot achieve # a similar speed to the fastest one in its group, this means this task # is actually not similar to other tasks in its group. # So we will remove it from its original group. if cur_flops < best_group_flops / self.beta and self.task_cts[task_idx] > 5 + max( self.task_cts[j] for j in group_ids if j != task_idx ): self.task_tags[task_idx] = None group_ids.remove(task_idx) def _restore_status(self, log_file, num_measures_per_round): """restore task_cts and best_costs from a log file""" str_target = str(self.tasks[0].target) workload_key_to_task_id = {t.workload_key: i for i, t in enumerate(self.tasks)} total_ct = -1 for total_ct, (inp, res) in enumerate(RecordReader(log_file)): if str(inp.task.target) != str_target: continue task_idx = workload_key_to_task_id.get(inp.task.workload_key, None) if task_idx is None: continue self.task_cts[task_idx] += 1 if res.error_no == 0: cost = array_mean(res.costs) if cost < self.best_costs[task_idx]: self.best_costs[task_idx] = cost self.task_best_cts[task_idx] = self.task_cts[task_idx] for idx in range(len(self.tasks)): if self.task_cts[idx] - self.task_best_cts[idx] > self.early_stopping_task: self.dead_tasks.add(idx) # The computation of taks_cts is just an estimation. # The estimation may not be accurate if the log file is changed externally or # `num_measures_per_round` is different from the last tuning. self.task_cts[idx] = int(self.task_cts[idx] / num_measures_per_round + 0.5) self.task_best_cts[idx] = int(self.task_best_cts[idx] / num_measures_per_round + 0.5) self.task_costs_history[idx].append(self.best_costs[idx]) self.cur_score = self._compute_score(self.best_costs) logger.info("TaskScheduler: Loaded %d measurement records from %s", total_ct + 1, log_file) class TaskSchedulerCallback: """The base class of task scheduler callback functions.""" def pre_tune(self, task_scheduler, task_id): """The callback before tuning each task. Parameters ---------- task_scheduler: TaskScheduler The task scheduler. task_id: int The task ID going to be tuned. """ # Do nothing by default def post_tune(self, task_scheduler, task_id): """The callback after tuning each task. Parameters ---------- task_scheduler: TaskScheduler The task scheduler. task_id: int The task ID be tuned. """ # Do nothing by default class PrintTableInfo(TaskSchedulerCallback): """The callback that prints a table of current progress.""" def pre_tune(self, task_scheduler, task_id): if task_scheduler.tune_option.verbose < 1: return _ffi_api.PrintTitle("Task Scheduler") print( "| ID " "| Task Description " "| Latency (ms) | Speed (GFLOPS) | Trials |" ) print( "----------------------------------------------------------------" "-------------------------------------------------" ) # content for i in range(len(task_scheduler.tasks)): id_str = f"{i}" latency_str = ( "%.3f" % (1e3 * task_scheduler.best_costs[i]) if task_scheduler.best_costs[i] < 1e9 else "-" ) task_desc = task_scheduler.tasks[i].desc speed_str = ( "%.2f" % (task_scheduler.tasks[i].compute_dag.flop_ct / task_scheduler.best_costs[i] / 1e9) if task_scheduler.best_costs[i] < 1e9 else "-" ) trials_str = "%d" % (task_scheduler.task_cts[i] * task_scheduler.num_measures_per_round) print( "| %4s | %61s | %12s | % 14s | %6s |" % (id_str, task_desc, latency_str, speed_str, trials_str) ) print( "----------------------------------------------------------------" "-------------------------------------------------" ) # overall info if all(cost < 1e9 for cost in task_scheduler.best_costs): total_latency_str = "%.3f" % (task_scheduler.cur_score * 1e3) else: total_latency_str = "-" print( "Estimated total latency: %s ms\tTrials: %d\tUsed time : %.0f s\tNext ID: %d\t" % (total_latency_str, task_scheduler.ct, time.time() - task_scheduler.tic, task_id) ) class LogEstimatedLatency(TaskSchedulerCallback): """Log the estimated latency to the file after tuning a task. Parameters ---------- log_file: str The log file path. """ def __init__(self, log_file): if os.path.exists(log_file): # Remove existing log os.remove(log_file) self.log_file = log_file def post_tune(self, task_scheduler, task_id): if all(cost < 1e9 for cost in task_scheduler.best_costs): total_latency_str = "%.3f" % (task_scheduler.cur_score * 1e3) else: total_latency_str = "N/A" with open(self.log_file, "a") as filep: filep.write( "ElapsedTime(s)\t%.0f\tEstimatedLatency(ms)\t%s\tTrials\t%d\n" % (time.time() - task_scheduler.tic, total_latency_str, task_scheduler.ct) ) filep.flush()
25,378
37.865237
100
py
tvm
tvm-main/python/tvm/auto_scheduler/utils.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=invalid-name """ Common utilities for auto_scheduler. """ from typing import Hashable import json import signal import threading import traceback import os import numpy as np try: import psutil except ImportError: psutil = None import tvm from tvm import rpc from tvm.tir import expr from tvm.tir.transform import Simplify from tvm.ir.transform import Sequential from ..te import Tensor, placeholder def decode_workload_key(workload_key): """Decode the workload key from a string to the name and arguments. The wokrload key is expected to be a list of "[func_name/hash, args ...]" in a JSON string. If not, then simply return the workload key as the name without arguments. Parameters ---------- workload_key: str The workload key in string. Format: "[func_name/hash, args ...]". Returns ------- name: str The workload function name or the DAG hash. args: Optional[Tuple[Any, ...]] The flatten arguments in a tuple, or None if the workload key format is not decodeable. """ def flatten_list(inp): ret = [] for elt in inp: if isinstance(elt, list): ret += flatten_list(elt) else: ret.append(elt) return ret try: key_list = json.loads(workload_key) if isinstance(key_list, list) and len(key_list) >= 1: return key_list[0], tuple(flatten_list(key_list[1:])) except json.decoder.JSONDecodeError: pass return workload_key, None def calc_workload_dis_factor(target_workload_pair, workload_pair): """Calculate the distance factor of the workload to the target workload. If two workloads are not compatible at all (i.e., different compute DAG or function), then the distance factor is "inf". Otherwise, we calculate the factor by traversing the workload arguments, which are the arguments of the compute function, or the output shapes for the ComputeDAG. The factor is calculated by the following rules: 1. For non-zero integer values: `product(target_arg / candidate_arg)`. 2. For non-integer or zero values: "inf" if not equal else 1. As a result, factor=1 is the optimal when two workloads are identical. Parameters ---------- target_workload_pair: Tuple[str, Optional[Tuple[Any, ...]]] The target workload pair: (hash, argument tuple). workload_pair: Tuple[str, Optional[Tuple[Any, ...]]] The candidate workload pair: (hash, argument tuple). Returns ------- dis_f: float The distance factor. """ target_key, target_args = target_workload_pair target_args = target_args if target_args is not None else [] key, args = workload_pair args = args if args is not None else [] # Not even the same func/DAG. if key != target_key or len(target_args) != len(args): return float("inf") dis_f = 1 for target_arg, arg in zip(target_args, args): if isinstance(target_arg, int): if target_arg == 0 or arg == 0: if target_arg != arg: return float("inf") elif target_arg % arg != 0: return float("inf") else: dis_f *= target_arg / arg elif target_arg != arg: return float("inf") return dis_f def get_func_name(func): """Get name of a function. Parameters ---------- func: Function The input function. Returns ------- name: str The function name. """ return func.func_name if hasattr(func, "func_name") else func.__qualname__ def get_const_int(exp): """Verifies expr is integer and get the constant value. Parameters ---------- exp : Union[tvm.tir.expr, int] The input expression. Returns ------- out_value : int The output. """ if isinstance(exp, int): return exp if not isinstance(exp, expr.IntImm): opt = Sequential([Simplify()]) exp = opt(exp) if not isinstance(exp, expr.IntImm): raise ValueError("Expect value to be constant int") return exp.value def get_const_tuple(in_tuple): """Verifies input tuple is IntImm, returns tuple of int. Parameters ---------- in_tuple : Tuple[tvm.tir.expr] The input. Returns ------- out_tuple : Tuple[Union[int,tvm.tir.Var,tvm.tir.Any]] The output tuple of int. The dynamic shape variables (Var or Any) will be preserved. """ ret = [] for elem in in_tuple: if isinstance(elem, (tvm.tir.Var, tvm.tir.expr.Any)): ret.append(elem) else: ret.append(get_const_int(elem)) return tuple(ret) def list_to_tuple(x): """Convert a list to a tuple recursively.""" assert isinstance(x, list) return tuple(list_to_tuple(y) if isinstance(y, list) else y for y in x) def serialize_args(args): """ Serialize arguments of a function to a hashable and jsonable tuple. Currently this is mainly used for tvm.tensor.Tensor """ ret = [] if args is None: return tuple(ret) for t in args: if isinstance(t, Tensor): t = ("TENSOR", get_const_tuple(t.shape), t.dtype) elif isinstance(t, list): t = list_to_tuple(t) assert isinstance(t, Hashable), str(t) + " is not hashable" ret.append(t) return tuple(ret) def deserialize_args(args): """The inverse function of :code:`serialize_args`""" ret = [] for t in args: if isinstance(t, (tuple, list)) and t[0] == "TENSOR": ret.append(placeholder(shape=t[1], dtype=t[2])) else: ret.append(t) return ret def kill_child_processes(parent_pid, sig=signal.SIGTERM): """kill all child processes recursively""" if not psutil: raise ImportError("psutil not found, try `pip install psutil` to fix this") try: parent = psutil.Process(parent_pid) except psutil.NoSuchProcess: return try: children = parent.children(recursive=True) for process in children: process.send_signal(sig) except psutil.NoSuchProcess: return # The maximum length of traceback information MAX_TRACEBACK_INFO_LEN = 512 def make_traceback_info(): """Get the error message from traceback.""" info = str(traceback.format_exc()) if len(info) > MAX_TRACEBACK_INFO_LEN: info = ( info[: MAX_TRACEBACK_INFO_LEN // 2] + "\n...\n" + info[-MAX_TRACEBACK_INFO_LEN // 2 :] ) return info class PropagatingThread(threading.Thread): """A thread that propagates the exception to the main thread""" def run(self): self.exc = None try: self.ret = self._target(*self._args, **self._kwargs) except Exception as e: # pylint: disable=broad-except self.exc = e def join(self, timeout=None): super(PropagatingThread, self).join(timeout) if self.exc: raise self.exc return self.ret def call_func_with_thread(func, args, kwargs): """Call a function within a new thread""" res = [] def wrapper(): res.append(func(*args, **kwargs)) t = PropagatingThread(target=wrapper) t.start() t.join() return res[0] def call_func_with_timeout( worker, timeout, func, args=(), kwargs=None ): # pylint: disable=unused-argument """Call a function with timeout""" worker.send(func, args, kwargs, timeout) try: res = worker.recv() except Exception: # pylint: disable=broad-except res = Exception(make_traceback_info()) return res def request_remote(device_key, host=None, port=None, priority=1, timeout=60): """Request a remote session. Parameters ---------- device_key : str The device key of registered device in tracker. host : Optional[str] The host address of rpc tracker. If is none, will use environment variable "TVM_TRACKER_HOST". port : Optional[int] The port of rpc tracker. If is none, will use environment variable "TVM_TRACKER_PORT". priority : int = 1 The priority of this request, larger is more prior. timeout : int = 60 The timeout of this session in second. Returns ------- remote : RPCSession The connected remote RPCSession. """ # connect to the tracker host = host or os.environ["TVM_TRACKER_HOST"] port = port or int(os.environ["TVM_TRACKER_PORT"]) tracker = rpc.connect_tracker(host, port) remote = tracker.request(device_key, priority=priority, session_timeout=timeout) return remote def check_remote(device_key, host=None, port=None, priority=100, timeout=10): """ Check the availability of a remote device. Parameters ---------- device_key: str device key of registered device in tracker. host: Optional[str] The host address of rpc tracker. If is none, will use environment variable "TVM_TRACKER_HOST". port: Optional[int] The port address of rpc tracker. If is none, will use environment variable "TVM_TRACKER_PORT". priority: int = 100 The priority of this request, larger is more prior. timeout: int = 10 The timeout of this check in seconds. Returns ------- available: bool True if can find available device. """ def _check(): request_remote(device_key, host, port, priority) t = threading.Thread(target=_check) t.start() t.join(timeout) return not t.is_alive() def array_mean(arr): """Compute mean of the elments in a TVM Array<PrimExpr> Parameters ---------- arr: Array A TVM Array<PrimExpr> Returns ------- mean: float The mean of the elements in the array """ return sum(x.value for x in arr) / len(arr) def to_str_round(x, decimal=6): """Convert an object to str and round float numbers Parameters ---------- x: Union[str, list, int, float, np.ndarray] The input object decimal: int The precision of decimal fraction Returns ------- ret: str The string format of these objects """ if isinstance(x, str): return x if isinstance(x, (list, tuple, np.ndarray)): return "[" + ", ".join([to_str_round(y, decimal=decimal) for y in x]) + "]" if isinstance(x, dict): return str({k: to_str_round(v) for k, v in x.items()}) if isinstance(x, int): return str(x) if isinstance(x, (np.float32, np.float64, float)): format_str = f"%.{decimal}f" return format_str % x raise ValueError(f"Invalid value: {str(x)}\ttype: {type(x)}")
11,629
27.296837
98
py
tvm
tvm-main/python/tvm/auto_scheduler/dispatcher.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The global context that dispatches best schedules to workloads. In auto-scheduler, a state (loop_state.py::StateObject) saves the schedule configuration by its transform_steps, so a state is used as a schedule configuration here. """ # pylint: disable=invalid-name import logging import pathlib from collections.abc import Iterable import numpy as np from tvm.contrib.utils import tempdir from tvm.tir.expr import FloatImm from .cost_model import RandomModel, XGBModel from .measure import LocalRPCMeasureContext from .measure_record import RecordToFile, load_records from .search_policy import PreloadMeasuredStates, SketchPolicy from .search_task import SearchTask, TuningOptions from .utils import calc_workload_dis_factor, decode_workload_key logger = logging.getLogger("auto_scheduler") class DispatchContext(object): """ Base class of dispatch context. """ current = None def __init__(self): self._old_ctx = DispatchContext.current def query(self, target, workload_key, has_complex_op, dag, func_name): """ Query the context to get the specific config for a workload. If this function cannot find the result inside this context, it will query the result from the upper contexts. Parameters ---------- target: Target The current target workload_key : str The workload key has_complex_op: bool Whether this workload has at least one complex op. dag: ComputeDAG The ComputeDAG of the workload. func_name: str The function name of this workload. Returns ------- state : StateObject The state that stores schedule configuration for the workload """ ret = self._query_inside(target, workload_key, func_name) if ret is None: ret = self._old_ctx.query(target, workload_key, has_complex_op, dag, func_name) return ret def update(self, target, workload_key, state): """ Update the config for a workload Parameters ---------- target: Target The current target workload_key : str The current workload_key. state : StateObject The state that stores schedule configuration for the workload """ raise NotImplementedError() def _query_inside(self, target, workload_key, func_name): """ Query the context to get the specific config for a workload. This function only query config inside this context. Parameters ---------- target: Target The current target workload_key : str The current workload_key. func_name: str The function name of this workload. Returns ------- state : StateObject The schedule configuration for the workload """ raise NotImplementedError() def __enter__(self): self._old_ctx = DispatchContext.current DispatchContext.current = self return self def __exit__(self, ptype, value, trace): DispatchContext.current = self._old_ctx class ApplyHistoryBest(DispatchContext): """ Apply the history best config Parameters ---------- records : str, list of str, or iterator of (auto_scheduler.measure.MeasureInput,\ auto_scheduler.measure.MeasureResult) Collection of tuning records. If is str, then it should be the filename of a records log file. Each row of this file is an encoded record pair. If it is an iterator, it can either be a set of str filenames which will be applied jointly, or a set of (input, result) tuples. n_lines: Optional[int] if it is not None, only load the first `n_lines` lines of log. include_compatible: bool When set to True, compatible records will also be considered. """ def __init__(self, records, n_lines=None, include_compatible=False): super(ApplyHistoryBest, self).__init__() self.include_compatible = include_compatible # Dict[str (target key), # Dict[str (workload hash), # Dict[tuple (workload args), tuple (State, cost)]]] self.best_by_targetkey = {} self.best_by_model = {} self._best_user_defined = {} self.load(records, n_lines) @staticmethod def get_workload_entry(best_records, target_key, workload_key): """Get the entry of the target key and workload key hash in the given best record map. Parameters ---------- best_records: Dict[str, Dict[str, Dict[str, Any]]] The best record map. target_key: str The first key to the best_records. workload_key: str The workload key that can be decoded to workload hash and args. Returns ------- entry: Dict[str, Any] The entry in best_records with target key and workload hash. workload_hash: str The workload hash decoded from workload_key. workload_args: Tuple[Any, ...] The hashable tuple of workload args decoded from workload_key. """ workload_hash, workload_args = decode_workload_key(workload_key) if target_key not in best_records: best_records[target_key] = {} if workload_hash not in best_records[target_key]: best_records[target_key][workload_hash] = {} return best_records[target_key][workload_hash], workload_hash, workload_args def load(self, records, n_lines=None): """Load records to this dispatch context Parameters ---------- records : str or iterator of (auto_scheduler.measure.MeasureInput,\ auto_scheduler.measure.MeasureResult) Collection of tuning records. If is str, then it should be the filename of a records log file. Each row of this file is an encoded record pair. Otherwise, it is an iterator. n_lines: Optional[int] if it is not None, only load the first `n_lines` lines of log """ joint_records = [] if not isinstance(records, Iterable) or isinstance(records, str): records = [records] for rec in records: if isinstance(rec, pathlib.Path): rec = str(rec) if isinstance(rec, str): rec = load_records(rec) joint_records += rec else: if rec is not None: joint_records.append(rec) if not joint_records: return best_by_targetkey = self.best_by_targetkey best_by_model = self.best_by_model counter = 0 for inp, res in joint_records: if n_lines is not None and counter >= n_lines: break counter += 1 if res.error_no != 0: continue costs = [x.value for x in res.costs if isinstance(x, FloatImm)] cost = np.mean(costs) # use target keys in tvm target system as key to build best map for k in inp.task.target.keys: entry, _, workload_args = self.get_workload_entry( best_by_targetkey, k, inp.task.workload_key ) if workload_args not in entry: entry[workload_args] = (inp.state, cost) else: _, other_cost = entry[workload_args] if other_cost > cost: entry[workload_args] = (inp.state, cost) # use model as key to build best map entry, _, workload_args = self.get_workload_entry( best_by_model, inp.task.target.model, inp.task.workload_key ) if workload_args not in entry: if inp.task.target.model != "unknown": entry[workload_args] = (inp.state, cost) else: _, other_cost = entry[workload_args] if other_cost > cost: entry[workload_args] = (inp.state, cost) logger.debug("Finish loading %d records", counter) def _query_inside(self, target, workload_key, func_name): if target is None: raise RuntimeError( "Need a target context to find the history best. " "Hint: If your target is llvm, use `with tvm.target.create('llvm'):`" " above the dispatcher call. So does other target. " ) def match_record(best_records, target_key, workload_key): """The helper function to match the record in the given map and return the matched state, or None if no match. """ ret = None entry, workload_hash, workload_args = self.get_workload_entry( best_records, target_key, workload_key ) if workload_args in entry: ret = entry[workload_args][0] elif self.include_compatible: best_cost = float("inf") for args, val in entry.items(): dis_f = calc_workload_dis_factor( (workload_hash, workload_args), (workload_hash, args) ) if dis_f == float("inf"): continue state, cost = val cost *= dis_f if ret is None or cost < best_cost: best_cost = cost ret = state return ret # first try matching by model ret = match_record(self._best_user_defined, target.model, workload_key) if ret is not None: return ret ret = match_record(self.best_by_model, target.model, workload_key) if ret is not None: return ret # then try matching by target key for k in target.keys: ret = match_record(self._best_user_defined, k, workload_key) if ret is not None: return ret ret = match_record(self.best_by_targetkey, k, workload_key) if ret is not None: return ret return None def update(self, target, workload_key, state): entry, _, workload_args = self.get_workload_entry( self._best_user_defined, target.model, workload_key ) entry[workload_args] = (state, 1) for k in target.keys: entry, _, _ = self.get_workload_entry(self._best_user_defined, k, workload_key) entry[workload_args] = (state, 1) class ApplyHistoryBestOrSample(ApplyHistoryBest): """ Apply the history best config, or sample a valid schedule if no config is found. Parameters ---------- records : str or iterator of (auto_scheduler.measure.MeasureInput,\ auto_scheduler.measure.MeasureResult) Collection of tuning records. If is str, then it should be the filename of a records log file. Each row of this file is an encoded record pair. Otherwise, it is an iterator. sample_simple_workloads: bool When False, sampling will not apply to simple workloads (w/o reduction). cost_model_file: str The filename of the pre-trained XGBoost cost model. If not present, then random model will be used. num_measure: int Meausre the top-N rank of sampled schedules on the device. The default -1 means no measurement and simply return the top-1 schedule ranked by the cost model. """ def __init__( self, records, sample_simple_workloads=False, cost_model_file=None, num_measure=-1 ): self.sample_simple_workloads = sample_simple_workloads self.num_measure = num_measure self.log_dir = tempdir() if cost_model_file is None: self.cost_model = RandomModel() else: self.cost_model = XGBModel() self.cost_model.load(cost_model_file) super(ApplyHistoryBestOrSample, self).__init__( records, n_lines=None, include_compatible=True ) def query(self, target, workload_key, has_complex_op, dag, func_name): if has_complex_op or self.sample_simple_workloads: ret = self._query_inside(target, workload_key, func_name) else: ret = super(ApplyHistoryBestOrSample, self)._query_inside( target, workload_key, func_name ) if ret is None: ret = self._old_ctx.query(target, workload_key, has_complex_op, dag, func_name) return ret def _query_inside(self, target, workload_key, func_name): ret = super(ApplyHistoryBestOrSample, self)._query_inside(target, workload_key, func_name) if ret is not None: return ret # Sampling valid schedules when no existing records can be used. task = SearchTask(workload_key=workload_key, target=target) measure_ctx = LocalRPCMeasureContext(min_repeat_ms=300) log_file = self.log_dir.relpath(f"{decode_workload_key(workload_key)[0]}.log") while ret is None: tune_option = TuningOptions( num_measure_trials=self.num_measure, runner=measure_ctx.runner, measure_callbacks=[RecordToFile(log_file)], verbose=0, ) search_policy = SketchPolicy( task, self.cost_model, params={ "eps_greedy": 0.01, "sample_init_min_population": 64, "evolutionary_search_num_iters": 0, }, init_search_callbacks=[PreloadMeasuredStates(log_file)], verbose=0, ) task.tune(tune_option, search_policy) # Load the sampled records and query again. self.load(log_file) ret = super(ApplyHistoryBestOrSample, self)._query_inside( target, workload_key, func_name ) del measure_ctx return ret class FallbackContext(DispatchContext): """ A fallback dispatch context. This is used as the root context. """ def __init__(self): super(FallbackContext, self).__init__() self.memory = {} # Verbose level: # 0: Completely silent. # 1: Warning the missing configs for querying complex tasks. # 2: Warning the missing configs for querying all tasks. self.verbose = 1 # a set to prevent print duplicated message self.messages = set() def query(self, target, workload_key, has_complex_op, dag, func_name): key = (str(target), workload_key) if key in self.memory: return self.memory[key] if self.verbose == 2 or (has_complex_op and self.verbose == 1): msg = ( f"-----------------------------------\n" f"{func_name}\n" f"Cannot find tuned schedules for target={target}, workload_key={workload_key}. " f"A fallback TOPI schedule is used, " f"which may bring great performance regression or even compilation failure. " f"Compute DAG info:\n{dag}" ) if msg not in self.messages: self.messages.add(msg) logger.warning(msg) state = None # cache this config to avoid duplicated warning message self.memory[key] = state return state def _query_inside(self, target, workload_key, func_name): _ = target = workload_key = func_name raise RuntimeError("This function should never be called") def update(self, target, workload_key, state): key = (str(target), workload_key) self.memory[key] = state DispatchContext.current = FallbackContext()
16,965
35.252137
98
py
tvm
tvm-main/python/tvm/auto_scheduler/loop_state.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=unused-import """ The definition of the "state" in the search. Each LoopState corresponds to a schedule for its ComputeDAG. A LoopState consists of: 1. a current loop structure; 2. a list of transformation steps used to construct the loop structure. The loop structure keeps a preview of how the schedule will finally look like after lowering the current state (e.g. number of iterators, the extent of each iterator, the compute_at locations ...). During the schedule search process, the loop structure can provide search policy with necessary information on how to manipulate the current state. The transform history is a sequence of `TransformStep` which will finally be mapped to TVM schedule primitives. The steps are also used for the serialization of a state. The LoopState can be seen as a lightweight loop structure IR specifically for schedule search. We don't use the existing TVM IR but to extend a new structure on it is because: 1. We want fast incremental change to the loop structures. The search policy needs to get the immediate loop structures update rather than after TVM lowering; 2. We want serializable transform history for replay, backtracking, and mutation; 3. We may create some macro schedule primitives that represent the combination of several TVM schedule primitives. When the search is finished, we will lower the state to TVM IR with TVM's schedule primitives. Since we share a lot of common objects during search, the transformation is implemented in copy on write style. All objects are immutable, which is similar to TVM IR. """ import tvm._ffi from tvm.te.tensor import Operation, Tensor from tvm.runtime import Object from . import _ffi_api @tvm._ffi.register_object("auto_scheduler.Iterator") class Iterator(Object): """A loop iterator structure.""" @tvm._ffi.register_object("auto_scheduler.Stage") class Stage(Object): """A stage in the compute declaration. Similar to tvm.te.schedule.Stage.""" # Static trans table for compute_at location # This is used to transform the compute_at location to C++ enum COMPUTE_AT_TRANS_TABLE = {"root": 0, "inlined": 1, "iter": 2} @tvm._ffi.register_object("auto_scheduler.State") class StateObject(Object): """The internal State object""" def __eq__(self, other): return _ffi_api.StateEqual(self, other) class State: """ A state in the search process. It consists of the current loop structure and a list of transformation steps used to construct it. Each State corresponds to a specific schedule for its ComputeDAG. Parameters ---------- state_object : StateObject The StateObject corresponding to C++ internal State object. dag : ComputeDAG The original ComputeDAG of this State. Notes ----- This is a wrapper class of StateObject to deal with copy-on-write property """ # Static trans table for thread bind and annotation # This is used to transform the annotation name to C++ enum ANNOTATION_TRANS_TABLE = { "none": 0, "unroll": 1, "vectorize": 2, "parallel": 3, "vthread": 4, "blockIdx.x": 5, "threadIdx.x": 6, "blockIdx.y": 7, "threadIdx.y": 8, "blockIdx.z": 9, "threadIdx.z": 10, "tensorize": 11, } def __init__(self, state_object, dag): self.state_object = state_object self.compute_dag = dag self.stage_id_map = {} # A dict maps operation to stage id self._update_stage_id_map() @property def stages(self): """ Returns ------- stages : List[Stage] """ return self.state_object.stages @property def transform_steps(self): """ Returns ------- transform_steps : List[transform_steps] """ return self.state_object.transform_steps @property def stage_ops(self): """ Returns ------- ops: List[Operation] """ return [stage.op for stage in self.stages] def bind(self, stage, iterator, thread_name): """Schedule primitive corresponding to `te.Stage.bind`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be binded, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be binded. thread_name : str The thread type to be binded. Candidates: - vthread - blockIdx.x - threadIdx.x - blockIdx.y - threadIdx.y - blockIdx.z - threadIdx.z Returns ------- res_it : Iterator The binded Iterator. """ if not thread_name in State.ANNOTATION_TRANS_TABLE.keys(): raise ValueError("Invalid thread_name: ", thread_name) self.state_object, res = _ffi_api.StateBind( self.state_object, self._resolve_stage_id(stage), iterator, State.ANNOTATION_TRANS_TABLE[thread_name], ) return res def parallel(self, stage, iterator): """Schedule primitive corresponding to `te.Stage.parallel`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be paralleled, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be paralleled. Returns ------- res_it : Iterator The paralleled Iterator. """ self.state_object, res = _ffi_api.StateParallel( self.state_object, self._resolve_stage_id(stage), iterator ) return res def unroll(self, stage, iterator, max_unroll=None): """Schedule primitive corresponding to `te.Stage.unroll`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be unrolled, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be unrolled. max_unroll : Optional[int] The max unroll limit. Iterator with extent larger than this limit will be skipped. Returns ------- res_it : Iterator The unrolled Iterator. """ self.state_object, res = _ffi_api.StateUnroll( self.state_object, self._resolve_stage_id(stage), iterator, max_unroll if max_unroll else -1, ) return res def vectorize(self, stage, iterator): """Schedule primitive corresponding to `te.Stage.vectorize`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be vectorized, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be vectorized. Returns ------- res_it : Iterator The vectorized Iterator. """ self.state_object, res = _ffi_api.StateVectorize( self.state_object, self._resolve_stage_id(stage), iterator ) return res def fuse(self, stage, iters): """Schedule primitive corresponding to `te.Stage.fuse`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be fused, which can be specified by the integer index, Operation, or output tensor of the stage. iters : List[Iterator] The iterators to be fused. Returns ------- res_it : Iterator The fused Iterator. Notes ----- If the iterators to be fused have stages attached at them(by compute_at), the fused result will become the new attach point. """ self.state_object, res = _ffi_api.StateFuse( self.state_object, self._resolve_stage_id(stage), iters ) return res def pragma(self, stage, iterator, pragma_type): """Schedule primitive corresponding to `te.Stage.pragma`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to add pragma, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to add pragma. pragma_type : str The pragma string. """ self.state_object = _ffi_api.StatePragma( self.state_object, self._resolve_stage_id(stage), iterator, pragma_type ) def reorder(self, stage, order): """Schedule primitive corresponding to `te.Stage.reorder`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be reordered, which can be specified by the integer index, Operation, or output tensor of the stage. order : List[Iterator] Iterators in the expected order. """ self.state_object = _ffi_api.StateReorder( self.state_object, self._resolve_stage_id(stage), order ) def split(self, stage, iterator, lengths, inner_to_outer=True): """Schedule primitive corresponding to `te.Stage.split`. See also the `te.Stage` for more details. This API supports multiple split factors. (e.g. with 2 split factors, the original iterator will be split to 3 parts, use `inner_to_outer` to control the split order) Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be split, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be split. lengths: List[int] The multiple split factors. Can be None to be filled by search policy. inner_to_outer: boolean = True Whether the factor go from inner to outer, or from outer to inner. Returns ------- res_its : List[Iterator] The splitted new Iterators. Notes ----- If we do split on an iterator which has stages attached at it(by compute_at), the inner most iterator of split results will become the new attach point. """ self.state_object, res = _ffi_api.StateSplit( self.state_object, self._resolve_stage_id(stage), iterator, lengths, inner_to_outer ) return res def follow_split(self, stage, iterator, src_step_id, n_split): """The schedule primitive similar to split, but uses split factors from previous steps. This step splits the iterator by the same factors as the given SplitStep. Notes ------ This step is useful in a scenario that we have subgraph Dense -> Relu, and we want to compute the Dense stage at ReLU. In this case, we need them to have the same tiling structure of common outer loops. The follow_split step could be used here to split the Dense stage and makes sure its splitting factors are the same as the given split step for the ReLU stage. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be split, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to split. src_step_id : int The index of the split step to be followed in the history. n_split : int The number of split level. Returns ------- res_its : List[Iterator] The splitted new Iterators. """ self.state_object, res = _ffi_api.StateFollowSplit( self.state_object, self._resolve_stage_id(stage), iterator, src_step_id, n_split ) return res def follow_fused_split(self, stage, iterator, src_step_ids, level, factor_or_nparts): """Schedule primitive extends to split step. This step is used to split an iterator by the same factors as the given list of SplitSteps and FuseSteps. Notes ------ This step is useful in a scenario that we have a subgraph in GPU schedule: Input -> Dense for i.0@j.0 = ... : Bind to blockIdx.x for i.1@j.1 = ... : Bind to threadIdx.x for i.2@j.2 = ... Input_shared = Input ... for k = ... Dense = ... We intend to apply cooperative fetching with the input stage, while the threadIdx.x axis is bound to an iterator generated by split & fuse step. The follow_fused_step is used split the iterator to 2 parts, while the split factor matches the final extent of the threadIdx.x bound iterator. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be split, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to split. src_step_ids : List[int] The indices of the split steps to be followed in the history. level : int Use the length in this split level. factor_or_nparts : bool True to use `factor` for split from inner to outer, False to use `nparts` for split from outer to inner. Returns ------- res_its : List[Iterator] The splitted new Iterators. """ self.state_object, res = _ffi_api.StateFollowFusedSplit( self.state_object, self._resolve_stage_id(stage), iterator, src_step_ids, level, factor_or_nparts, ) return res def storage_align(self, stage, iterator, factor, offset): """Schedule primitive corresponding to `te.Stage.storage_align`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be storage aligned, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The iterator to be aligned. factor : int The factor in alignment specification. offset : int The offset in the alignment specification. """ self.state_object = _ffi_api.StateStorageAlign( self.state_object, self._resolve_stage_id(stage), iterator, factor, offset ) def compute_at(self, stage, target_stage, target_iter): """Schedule primitive corresponding to `te.Stage.compute_at`. See also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The source Stage of computed at, which can be specified by the integer index, Operation, or output tensor of the stage. target_stage : Union[int, Operation, Tensor] The target stage of compute_at, which can be specified by the integer index, Operation, or output tensor of the stage. target_iter : Iterator The target Iterator of compute_at. Notes ----- After compute_at, we need careful dependency analysis to compute the accurate bound information. However, it is relatively expensive and complicated, so we just fill "None" as bound for the newly created iterators. Call ComputeDAG::InferBound on the returned state to get the complete bound information. """ self.state_object = _ffi_api.StateComputeAt( self.state_object, self._resolve_stage_id(stage), self._resolve_stage_id(target_stage), target_iter, ) def compute_inline(self, stage): """Schedule primitive corresponding to `te.Stage.compute_inline`, see also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be marked compute inlined, which can be specified by the integer index, Operation, or output tensor of the stage. """ self.state_object = _ffi_api.StateComputeInline( self.state_object, self._resolve_stage_id(stage) ) def compute_root(self, stage): """Schedule primitive corresponding to `te.Stage.compute_root`. Ssee also the `te.Stage` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be marked compute at root, which can be specified by the integer index, Operation, or output tensor of the stage. Notes ----- After compute_root, we need careful dependency analysis to compute the accurate bound information. However, it is relatively expensive and complicated, so we just fill "None" as bound for the newly created iterators. Call ComputeDAG::InferBound on the returned state to get the complete bound information. """ self.state_object = _ffi_api.StateComputeRoot( self.state_object, self._resolve_stage_id(stage) ) def cache_read(self, stage, scope_name, reader_stages): """Schedule primitive corresponding to `te.Schedule.cache_read`. See also the `te.Schedule` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be cache_read, which can be specified by the integer index, Operation, or output tensor of the stage. scope_name : str The scope name of the newly added read stage. reader_stages : List[Union[int, Operation, Tensor]] The reader stages. Each of the list can be specified by the integer index, Operation, or output tensor of the stage. Returns ------- new_stage_op : Operator The Operator of the new added stage. Notes ----- Cache read step will insert an extra stage to the original ComputeDAG (at the back of the target stage). """ reader_stage_ids = [self._resolve_stage_id(i) for i in reader_stages] self.state_object, new_stage_id = _ffi_api.StateCacheRead( self.state_object, self._resolve_stage_id(stage), scope_name, reader_stage_ids, self.compute_dag, ) # Add a new stage will change all ops behind the added stage. But we still want to keep the # original ops map, apply stage id offset to stage_id_map to make them work. self._apply_stage_id_offset(int(new_stage_id)) self._update_stage_id_map() return self.stages[int(new_stage_id)].op def cache_write(self, stage, scope_name): """Schedule primitive corresponding to `te.Schedule.cache_write`. See also the `te.Schedule` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be cache_write, which can be specified by the integer index, Operation, or output tensor of the stage. scope_name : str The scope name of the newly added compute stage. Returns ------- new_stage_op : Operator The Operator of the new added stage. Notes ----- Cache write step will insert an extra stage to the original ComputeDAG (in the front of the target stage). This step will cache write all output tensors of the target stage. """ self.state_object, new_stage_id = _ffi_api.StateCacheWrite( self.state_object, self._resolve_stage_id(stage), scope_name, self.compute_dag ) # Add a new stage will change all ops behind the added stage. But we still want to keep the # original ops map, apply stage id offset to stage_id_map to make them work. self._apply_stage_id_offset(int(new_stage_id)) self._update_stage_id_map() return self.stages[int(new_stage_id)].op def rfactor(self, stage, iterator, factor_iter_id): """Schedule primitive corresponding to `te.Schedule.rfactor`. See also the `te.Schedule` for more details. Parameters ---------- stage : Union[int, Operation, Tensor] The Stage to be factored, which can be specified by the integer index, Operation, or output tensor of the stage. iterator : Iterator The reduction iterator to be factored. factor_iter_id : int The position where the new iterator is placed. Returns ------- new_stage_op : Operator The Operator of the new added stage. Notes ----- Rfactor step will insert an extra stage to the original ComputeDAG (in the front of the target stage). """ self.state_object, new_stage_id = _ffi_api.StateRfactor( self.state_object, self._resolve_stage_id(stage), iterator, factor_iter_id, self.compute_dag, ) # Add a new stage will change all ops behind the added stage. But we still want to keep the # original ops map, apply stage id offset to stage_id_map to make them work. self._apply_stage_id_offset(int(new_stage_id)) self._update_stage_id_map() return self.stages[int(new_stage_id)].op def copy(self): """Do deep copy of this State.""" state = State(self.state_object, self.compute_dag) state.stage_id_map = self.stage_id_map.copy() return state def _resolve_stage_id(self, stage_id): if isinstance(stage_id, Operation): return self.stage_id_map[stage_id] if isinstance(stage_id, Tensor): return self.stage_id_map[stage_id.op] if isinstance(stage_id, int): return stage_id raise ValueError( "Invalid stage: " + stage_id + " . Expect to be a int, Operation or Tensor" ) def _update_stage_id_map(self): for index, stage in enumerate(self.stages): self.stage_id_map[stage.op] = index def _apply_stage_id_offset(self, start_id, offset=1): for key, value in self.stage_id_map.items(): if value >= start_id: self.stage_id_map[key] = value + offset def __getitem__(self, key): if isinstance(key, Tensor): key = key.op if isinstance(key, Operation): return self.stages[self.stage_id_map[key]] raise ValueError("Invalid item: " + key + " . Expect to be a Operation or Tensor") def __str__(self): return str(self.state_object) def __eq__(self, other): return _ffi_api.StateEqual(self.state_object, other.state_object)
24,583
38.71567
99
py
tvm
tvm-main/python/tvm/auto_scheduler/_ffi_api.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ Register FFI APIs from C++ for the namespace tvm.auto_scheduler. """ import tvm._ffi tvm._ffi._init_api("auto_scheduler", __name__)
924
39.217391
72
py
tvm
tvm-main/python/tvm/auto_scheduler/measure.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ Distributed measurement infrastructure to measure the runtime costs of tensor programs. These functions are responsible for building the tvm module, uploading it to remote devices, recording the running time costs, and checking the correctness of the output. We separate the measurement into two steps: build and run. A builder builds the executable binary files and a runner runs the binary files to get the measurement results. The flow of data structures is . `ProgramBuilder` `ProgramRunner` `MeasureInput` -----------------> `BuildResult` ----------------> `MeasureResult` We implement these in python to utilize python's multiprocessing and error handling. """ import logging import multiprocessing import os import shutil import tempfile import time import tvm._ffi from tvm.autotvm.env import AutotvmGlobalScope, reset_global_scope from tvm.contrib import ndk, tar from tvm.contrib.popen_pool import PopenPoolExecutor, PopenWorker, StatusKind from tvm.driver import build_module from tvm.ir import transform from tvm.runtime import Object, module, ndarray from tvm.target import Target from . import _ffi_api from .loop_state import StateObject from .utils import ( call_func_with_timeout, check_remote, get_const_tuple, get_func_name, make_traceback_info, request_remote, ) from .workload_registry import ( deserialize_workload_registry_entry, serialize_workload_registry_entry, ) # pylint: disable=invalid-name logger = logging.getLogger("auto_scheduler") # The time cost for measurements with errors # We use 1e10 instead of sys.float_info.max for better readability in log MAX_FLOAT = 1e10 class BuildFunc: """store build_func name and callable to class variable. name: str = "default" The name of registered build function. build_func: callable = tar.tar The callable of registered build function. """ name = "default" build_func = tar.tar @tvm._ffi.register_object("auto_scheduler.MeasureCallback") class MeasureCallback(Object): """The base class of measurement callback functions.""" @tvm._ffi.register_object("auto_scheduler.PythonBasedMeasureCallback") class PythonBasedMeasureCallback(MeasureCallback): """Base class for measure callbacks implemented in python""" def __init__(self): def callback_func(policy, inputs, results): self.callback(policy, inputs, results) self.__init_handle_by_constructor__(_ffi_api.PythonBasedMeasureCallback, callback_func) def callback(self, policy, inputs, results): """The callback function. Parameters ---------- policy: auto_scheduler.search_policy.SearchPolicy The search policy. inputs : List[auto_scheduler.measure.MeasureInput] The measurement inputs results : List[auto_scheduler.measure.MeasureResult] The measurement results """ raise NotImplementedError @tvm._ffi.register_object("auto_scheduler.MeasureInput") class MeasureInput(Object): """Store the input of a measurement. Parameters ---------- task : SearchTask The SearchTask of this measurement. state : Union[State, StateObject] The State to be measured. """ def __init__(self, task, state): state = state if isinstance(state, StateObject) else state.state_object self.__init_handle_by_constructor__(_ffi_api.MeasureInput, task, state) def serialize(self): """Custom serialization to workaround MeasureInput not exposing all its members to the TVM ffi interface. Note that we do not implement __getstate__ as it does not seem to work with initialization of the workload registry (maybe because of initialization order?). """ return [ _ffi_api.SerializeMeasureInput(self), serialize_workload_registry_entry(self.task.workload_key), ] @staticmethod def deserialize(data): inp = _ffi_api.DeserializeMeasureInput(data[0]) deserialize_workload_registry_entry(data[1]) return recover_measure_input(inp) @tvm._ffi.register_object("auto_scheduler.BuildResult") class BuildResult(Object): """Store the result of a build. Parameters ---------- filename : Optional[str] The filename of built binary file. args : List[Tensor] The arguments. error_no : int The error code. error_msg : Optional[str] The error message if there is any error. time_cost : float The time cost of build. """ def __init__(self, filename, args, error_no, error_msg, time_cost): filename = filename if filename else "" error_msg = error_msg if error_msg else "" self.__init_handle_by_constructor__( _ffi_api.BuildResult, filename, args, error_no, error_msg, time_cost ) @tvm._ffi.register_object("auto_scheduler.MeasureResult") class MeasureResult(Object): """Store the results of a measurement. Parameters ---------- costs : List[float] The time costs of execution. error_no : int The error code. error_msg : Optional[str] The error message if there is any error. all_cost : float The time cost of build and run. timestamp : float The time stamps of this measurement. """ def __init__(self, costs, error_no, error_msg, all_cost, timestamp): error_msg = error_msg if error_msg else "" self.__init_handle_by_constructor__( _ffi_api.MeasureResult, costs, error_no, error_msg, all_cost, timestamp ) def recover_measure_input(inp, rebuild_state=False): """ Recover a deserialized MeasureInput by rebuilding the missing fields. 1. Rebuid the compute_dag in inp.task 2. (Optional) Rebuild the stages in inp.state Parameters ---------- inp: MeasureInput The deserialized MeasureInput rebuild_state: bool = False Whether rebuild the stages in MeasureInput.State Returns ------- new_input: MeasureInput The fully recovered MeasureInput with all fields rebuilt. """ # pylint: disable=import-outside-toplevel from .search_task import SearchTask # lazily import to avoid recursive dependency task = inp.task task.target, task.target_host = Target.canon_target_and_host(task.target, task.target_host) new_task = SearchTask( workload_key=task.workload_key, target=task.target, hardware_params=task.hardware_params, layout_rewrite_option=task.layout_rewrite_option, task_inputs=list(task.task_input_names), ) if rebuild_state: new_state = new_task.compute_dag.infer_bound_from_state(inp.state) else: new_state = inp.state return MeasureInput(new_task, new_state) @tvm._ffi.register_object("auto_scheduler.ProgramBuilder") class ProgramBuilder(Object): """The base class of ProgramBuilders.""" def build(self, measure_inputs, verbose=1): """Build programs and return results. Parameters ---------- measure_inputs : List[MeasureInput] A List of MeasureInput. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during program building. Returns ------- res : List[BuildResult] """ return _ffi_api.ProgramBuilderBuild(self, measure_inputs, verbose) @tvm._ffi.register_object("auto_scheduler.ProgramRunner") class ProgramRunner(Object): """The base class of ProgramRunners.""" def run(self, measure_inputs, build_results, verbose=1): """Run measurement and return results. Parameters ---------- measure_inputs : List[MeasureInput] A List of MeasureInput. build_results : List[BuildResult] A List of BuildResult to be ran. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during program running. Returns ------- res : List[MeasureResult] """ return _ffi_api.ProgramRunnerRun(self, measure_inputs, build_results, verbose) @tvm._ffi.register_object("auto_scheduler.ProgramMeasurer") class ProgramMeasurer(Object): """ Measurer that measures the time costs of tvm programs This class combines ProgramBuilder and ProgramRunner, and provides a simpler API. Parameters ---------- builder : ProgramBuilder The ProgramBuilder to build programs runner : ProgramRunner The ProgramRunner to measure programs. callbacks : List[MeasureCallback] Callbacks to be called after each measurement batch verbose : int The Verbosity level: 0 for silent, 1 to output information during program max_continuous_error : Optional[int] The number of allowed maximum continuous error before stop the tuning """ def __init__(self, builder, runner, callbacks, verbose, max_continuous_error=None): max_continuous_error = max_continuous_error or -1 # -1 means using the default value self.__init_handle_by_constructor__( _ffi_api.ProgramMeasurer, builder, runner, callbacks, verbose, max_continuous_error ) @tvm._ffi.register_object("auto_scheduler.LocalBuilder") class LocalBuilder(ProgramBuilder): """LocalBuilder use local CPU cores to build programs in parallel. Parameters ---------- timeout : int = 15 The timeout limit (in second) for each build thread. This is used in a wrapper of the multiprocessing.Process.join(). n_parallel : int = multiprocessing.cpu_count() Number of threads used to build in parallel. build_func: callable or str = "default" If is 'default', use default build function If is 'ndk', use function for android ndk If is callable, use it as custom build function, expect lib_format field. """ def __init__(self, timeout=15, n_parallel=multiprocessing.cpu_count(), build_func="default"): if build_func == "default": BuildFunc.name = "default" BuildFunc.build_func = tar.tar elif build_func == "ndk": BuildFunc.name = "ndk" BuildFunc.build_func = ndk.create_shared elif callable(build_func): BuildFunc.name = "custom" BuildFunc.build_func = build_func else: raise ValueError("Invalid build_func" + build_func) self.__init_handle_by_constructor__( _ffi_api.LocalBuilder, timeout, n_parallel, BuildFunc.name ) @tvm._ffi.register_object("auto_scheduler.LocalRunner") class LocalRunner(ProgramRunner): """LocalRunner that uses local CPU/GPU to measures the time cost of programs. Parameters ---------- timeout : int = 10 The timeout limit (in second) for each run. This is used in a wrapper of the multiprocessing.Process.join(). number : int = 3 The number of times to run the generated code for taking average. We call these runs as one `repeat` of measurement. repeat : int = 1 The number of times to repeat the measurement. In total, the generated code will be run (1 + number x repeat) times, where the first "1" is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms : int = 100 The minimum duration of one `repeat` in milliseconds. By default, one `repeat` contains `number` runs. If this parameter is set, the parameters `number` will be dynamically adjusted to meet the minimum duration requirement of one `repeat`. i.e., When the run time of one `repeat` falls below this time, the `number` parameter will be automatically increased. cooldown_interval : float = 0.0 The cool down interval between two measurements in seconds. enable_cpu_cache_flush: bool = False Whether to flush cache on CPU between repeated measurements. Flushing cache can make the measured latency of one operator closer to its actual latency during end-to-end inference. To make this option effective, the argument `number` should also be set to 1. This is only has effect on CPU task. device: int = 0 Which device to run on if multiple are available. """ def __init__( self, timeout=10, number=3, repeat=1, min_repeat_ms=100, cooldown_interval=0.0, enable_cpu_cache_flush=False, device=0, ): if enable_cpu_cache_flush: number = 1 min_repeat_ms = 0 self.__init_handle_by_constructor__( _ffi_api.LocalRunner, timeout, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, device, ) @tvm._ffi.register_object("auto_scheduler.RPCRunner") class RPCRunner(ProgramRunner): """RPCRunner that uses RPC call to measures the time cost of programs on remote devices. Or sometime we may need to use RPC even in local running to insulate the thread environment. (e.g. running CUDA programs) Parameters ---------- key : str The key of the device registered in the RPC tracker. host : str The host address of the RPC Tracker. port : int The port of RPC Tracker. priority : int = 1 The priority of this run request, larger is more prior. n_parallel : int = 1 The number of tasks run in parallel. timeout : int = 10 The timeout limit (in second) for each run. This is used in a wrapper of the multiprocessing.Process.join(). number : int = 3 The number of times to run the generated code for taking average. We call these runs as one `repeat` of measurement. repeat : int = 1 The number of times to repeat the measurement. In total, the generated code will be run (1 + number x repeat) times, where the first "1" is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms : int = 100 The minimum duration of one `repeat` in milliseconds. By default, one `repeat` contains `number` runs. If this parameter is set, the parameters `number` will be dynamically adjusted to meet the minimum duration requirement of one `repeat`. i.e., When the run time of one `repeat` falls below this time, the `number` parameter will be automatically increased. cooldown_interval : float = 0.0 The cool down interval between two measurements in seconds. enable_cpu_cache_flush: bool = False Whether to flush cache on CPU between repeated measurements. Flushing cache can make the measured latency of one operator closer to its actual latency during end-to-end inference. To make this option effective, the argument `number` should also be set to 1. This is only has effect on CPU task. device: int = 0 Which device to run on if multiple are available. """ def __init__( self, key, host, port, priority=1, n_parallel=1, timeout=10, number=3, repeat=1, min_repeat_ms=100, cooldown_interval=0.0, enable_cpu_cache_flush=False, device=0, ): self.__init_handle_by_constructor__( _ffi_api.RPCRunner, key, host, port, priority, n_parallel, timeout, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, device, ) if check_remote(key, host, port, priority, timeout): print("Get devices for measurement successfully!") else: raise RuntimeError( "Cannot get remote devices from the tracker. " "Please check the status of tracker by " "'python -m tvm.exec.query_rpc_tracker --port [THE PORT YOU USE]' " "and make sure you have free devices on the queue status." ) class LocalRPCMeasureContext: """A context wrapper for running RPCRunner locally. This will launch a local RPC Tracker and local RPC Server. Parameters ---------- priority : int = 1 The priority of this run request, larger is more prior. n_parallel : int = 1 The number of tasks run in parallel. timeout : int = 10 The timeout limit (in second) for each run. This is used in a wrapper of the multiprocessing.Process.join(). number : int = 3 The number of times to run the generated code for taking average. We call these runs as one `repeat` of measurement. repeat : int = 1 The number of times to repeat the measurement. In total, the generated code will be run (1 + number x repeat) times, where the first "1" is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms : int = 0 The minimum duration of one `repeat` in milliseconds. By default, one `repeat` contains `number` runs. If this parameter is set, the parameters `number` will be dynamically adjusted to meet the minimum duration requirement of one `repeat`. i.e., When the run time of one `repeat` falls below this time, the `number` parameter will be automatically increased. cooldown_interval : float = 0.0 The cool down interval between two measurements in seconds. enable_cpu_cache_flush: bool = False Whether to flush cache on CPU between repeated measurements. Flushing cache can make the measured latency of one operator closer to its actual latency during end-to-end inference. To make this option effective, the argument `number` should also be set to 1. This is only has effect on CPU task. device: int = 0 Which device to run on if multiple are available. """ def __init__( self, priority=1, n_parallel=1, timeout=10, number=3, repeat=1, min_repeat_ms=0, cooldown_interval=0.0, enable_cpu_cache_flush=False, device=0, ): # pylint: disable=import-outside-toplevel from tvm.rpc.server import Server from tvm.rpc.tracker import Tracker self.tracker = Tracker(port=9000, port_end=10000, silent=True) device_key = f"$local$device${self.tracker.port}" self.server = Server( port=self.tracker.port, port_end=10000, key=device_key, silent=True, tracker_addr=("127.0.0.1", self.tracker.port), ) self.runner = RPCRunner( device_key, "127.0.0.1", self.tracker.port, priority, n_parallel, timeout, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, device, ) # Wait for the processes to start time.sleep(0.5) def __del__(self): # Close the tracker and server before exit self.tracker.terminate() self.server.terminate() time.sleep(0.5) class MeasureErrorNo(object): """Error type for MeasureResult.""" NO_ERROR = 0 # No error INSTANTIATION_ERROR = 1 # Errors happen when apply transform steps from init state COMPILE_HOST = 2 # Errors happen when compiling code on host (e.g., tvm.build) COMPILE_DEVICE = 3 # Errors happen when compiling code on device # (e.g. OpenCL JIT on the device) RUNTIME_DEVICE = 4 # Errors happen when run program on device WRONG_ANSWER = 5 # Answer is wrong when compared to a reference output BUILD_TIMEOUT = 6 # Timeout during compilation RUN_TIMEOUT = 7 # Timeout during run UNKNOWN_ERROR = 8 # Unknown error def _local_build_worker(inp_serialized, build_func, verbose): tic = time.time() inp = MeasureInput.deserialize(inp_serialized) task = inp.task task.target, task.target_host = Target.canon_target_and_host(task.target, task.target_host) error_no = MeasureErrorNo.NO_ERROR error_msg = None args = [] try: sch, args = task.compute_dag.apply_steps_from_state( inp.state, layout_rewrite=task.layout_rewrite_option ) # pylint: disable=broad-except except Exception: error_no = MeasureErrorNo.INSTANTIATION_ERROR error_msg = make_traceback_info() if error_no == 0: dirname = tempfile.mkdtemp() filename = os.path.join(dirname, "tmp_func." + build_func.output_format) try: with transform.PassContext().current(): func = build_module.build(sch, args, target=task.target) func.export_library(filename, build_func) # pylint: disable=broad-except except Exception: error_no = MeasureErrorNo.COMPILE_HOST error_msg = make_traceback_info() else: filename = "" if verbose >= 1: if error_no == MeasureErrorNo.NO_ERROR: print(".", end="", flush=True) else: print(".E", end="", flush=True) # Build error return filename, args, error_no, error_msg, time.time() - tic def local_build_worker(args): """ Build function of LocalBuilder to be ran in the Builder thread pool. Parameters ---------- args: Tuple[MeasureInput, callable, int] inputs, build-func, verbose args passed to local_builder_build Returns ------- res : BuildResult The build result of this Builder thread. """ inp, build_func, verbose = args return _local_build_worker(inp, build_func, verbose) @tvm._ffi.register_func("auto_scheduler.local_builder.build") def local_builder_build(inputs, timeout, n_parallel, build_func="default", verbose=1): """ Build function of LocalBuilder to build the MeasureInputs to runnable modules. Parameters ---------- inputs : List[MeasureInput] The MeasureInputs to be built. timeout : int The timeout limit (in second) for each build thread. This is used in a wrapper of the multiprocessing.Process.join(). n_parallel : int Number of threads used to build in parallel. build_func : str = 'default' The name of build function to process the built module. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during program building. Returns ------- res : List[BuildResult] The build results of these MeasureInputs. """ assert build_func == BuildFunc.name, ( "BuildFunc.name: " + BuildFunc.name + ", but args is: " + build_func ) executor = PopenPoolExecutor( n_parallel, timeout, reset_global_scope, (AutotvmGlobalScope.current,) ) tuple_res = executor.map_with_error_catching( local_build_worker, [(i.serialize(), BuildFunc.build_func, verbose) for i in inputs] ) results = [] for res in tuple_res: if res.status == StatusKind.COMPLETE: results.append(BuildResult(*res.value)) elif res.status == StatusKind.TIMEOUT: if verbose >= 1: print(".T", end="", flush=True) # Build timeout results.append(BuildResult(None, [], MeasureErrorNo.BUILD_TIMEOUT, None, timeout)) elif res.status == StatusKind.EXCEPTION: if verbose >= 1: print(".E", end="", flush=True) # Build error results.append( BuildResult(None, [], MeasureErrorNo.COMPILE_HOST, repr(res.value), timeout) ) else: raise ValueError("Result status is not expected. Unreachable branch") return results TASK_INPUT_CHECK_FUNC_REGISTRY = {} def register_task_input_check_func(func_name, f=None, override=False): """Register a function that checks the input buffer map. The input function should take a list of Tensor wich indicate the Input/output Tensor of a TVM subgraph and return a Map from the input Tensor to its buffer name. Parameters ---------- func_name : Union[Function, str] The check function that returns the compute declaration Tensors or its function name. f : Optional[Function] The check function to be registered. override : boolean = False Whether to override existing entry. Examples -------- .. code-block:: python @auto_scheduler.register_task_input_check_func def check_task_input_by_placeholder_name(args : List[Tensor]): tensor_input_map = {} for arg in args: if isinstance(arg.op, tvm.te.PlaceholderOp): if arg.op.name != "placeholder": tensor_input_map[arg] = arg.op.name return tensor_input_map """ global TASK_INPUT_CHECK_FUNC_REGISTRY if callable(func_name): f = func_name func_name = get_func_name(f) if not isinstance(func_name, str): raise ValueError("expect string function name") def register(myf): """internal register function""" if func_name in TASK_INPUT_CHECK_FUNC_REGISTRY and not override: raise RuntimeError(f"{func_name} has been registered already") TASK_INPUT_CHECK_FUNC_REGISTRY[func_name] = myf return myf if f: return register(f) return register def prepare_input_map(args, workload_key=None): """This function deals with special task inputs. Map the input Tensor of a TVM subgraph to a specific buffer name in the global buffer map. Parameters ---------- args : List[Tensor] Input/output Tensor of a TVM subgraph. workload_key: Optional[str] The workload for which these inputs are being prepared. This is used to identify if an input is being provided by (see `register_task_input_buffer`). Returns ------- Dict[Tensor, str] : Map from the input Tensor to its buffer name. Notes ----- The buffer name is specially designed, and these buffer should be provided in `SearchTask(..., task_inputs={...})`. """ # pylint: disable=import-outside-toplevel global TASK_INPUT_CHECK_FUNC_REGISTRY from .search_task import TASK_INPUT_BUFFER_TABLE # A dict that maps the input tensor arg to a buffer name tensor_input_map = {} # Case 0: Check placeholder name for arg in args: if isinstance(arg.op, tvm.te.PlaceholderOp): if ( workload_key and workload_key in TASK_INPUT_BUFFER_TABLE and arg.op.name in TASK_INPUT_BUFFER_TABLE[workload_key] ): tensor_input_map[arg] = arg.op.name # Case 1: Check specific tensor inputs for func_name in TASK_INPUT_CHECK_FUNC_REGISTRY: func = TASK_INPUT_CHECK_FUNC_REGISTRY[func_name] tensor_input_map.update(func(args)) return tensor_input_map def prepare_runner_args(inp, build_res): """This function prepares the pre-defined arguments in `TASK_INPUT_BUFFER_TABLE` for local/rpc runner in main process Parameters ---------- inp : MeasureInput Measure input to be measured. build_res : BuildResult Build result to be measured. Returns ------- List[Optional[numpy.ndarray]] : List of arguments for running the program. If the argument does not have a pre-defined input buffer, None is added to the list as a placeholder. """ # pylint: disable=import-outside-toplevel from .search_task import get_task_input_buffer # lazily import to avoid recursive dependency task_input_names = inp.task.task_input_names tensor_input_map = prepare_input_map(build_res.args, inp.task.workload_key) if not task_input_names: tensor_input_map = {} args = [] task_inputs_count = 0 for arg in build_res.args: if arg in tensor_input_map: tensor_name = tensor_input_map[arg] if tensor_name in task_input_names: task_input_buffer = get_task_input_buffer(inp.task.workload_key, tensor_name) # convert tvm.NDArray to picklable numpy.ndarray args.append(task_input_buffer.numpy()) task_inputs_count += 1 else: raise ValueError( f"{tensor_name} not found in task_inputs, " f"should provide with `SearchTask(..., task_inputs={{...}})`" ) else: args.append(None) if task_inputs_count != len(task_input_names): raise RuntimeError("task_inputs not fully matched, check if there's any unexpected error") return args def _timed_eval_func( inp_serialized, build_res, args, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, verbose, device, ): inp = MeasureInput.deserialize(inp_serialized) tic = time.time() error_no = 0 error_msg = None try: func = module.load_module(build_res.filename) dev = ndarray.device(str(inp.task.target), device) # Limitation: # We can not get PackFunction directly in the remote mode as it is wrapped # under the std::function. We could lift the restriction later once we fold # the PackedFunc as an object. Currently, we pass function name to work # around it. f_prepare = "cache_flush_cpu_non_first_arg" if enable_cpu_cache_flush else "" time_f = func.time_evaluator( func.entry_name, dev, number=number, repeat=repeat, min_repeat_ms=min_repeat_ms, f_preproc=f_prepare, ) # pylint: disable=broad-except except Exception: costs = (MAX_FLOAT,) error_no = MeasureErrorNo.COMPILE_DEVICE error_msg = make_traceback_info() if error_no == 0: try: random_fill = tvm.get_global_func("tvm.contrib.random.random_fill", True) assert random_fill, "Please make sure USE_RANDOM is ON in the config.cmake" assert len(args) == len(build_res.args) loc_args = [] # pylint: disable=consider-using-enumerate for idx in range(len(args)): if args[idx] is None: build_res_arg = build_res.args[idx] empty_array = ndarray.empty( get_const_tuple(build_res_arg.shape), build_res_arg.dtype, dev ) random_fill(empty_array) loc_args.append(empty_array) else: loc_args.append(ndarray.array(args[idx], dev)) dev.sync() costs = time_f(*loc_args).results # pylint: disable=broad-except except Exception: costs = (MAX_FLOAT,) error_no = MeasureErrorNo.RUNTIME_DEVICE error_msg = make_traceback_info() shutil.rmtree(os.path.dirname(build_res.filename)) toc = time.time() time.sleep(cooldown_interval) if verbose >= 1: if error_no == MeasureErrorNo.NO_ERROR: print("*", end="", flush=True) else: print("*E", end="", flush=True) # Run error return costs, error_no, error_msg, toc - tic + build_res.time_cost, toc @tvm._ffi.register_func("auto_scheduler.local_runner.run") def local_run( inputs, build_results, timeout=10, number=3, repeat=1, min_repeat_ms=0, cooldown_interval=0, enable_cpu_cache_flush=False, verbose=1, device=0, ): """ Run function of LocalRunner to test the performance of the input BuildResults. Parameters ---------- inputs : List[MeasureInput] The MeasureInputs to be measured. build_results : List[BuildResult] The BuildResults to be measured. timeout : int = 10 The timeout limit (in second) for each run. This is used in a wrapper of the multiprocessing.Process.join(). number : int = 3 The number of times to run the generated code for taking average. We call these runs as one `repeat` of measurement. repeat : int = 1 The number of times to repeat the measurement. In total, the generated code will be run (1 + number x repeat) times, where the first "1" is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms : int = 0 The minimum duration of one `repeat` in milliseconds. By default, one `repeat` contains `number` runs. If this parameter is set, the parameters `number` will be dynamically adjusted to meet the minimum duration requirement of one `repeat`. i.e., When the run time of one `repeat` falls below this time, the `number` parameter will be automatically increased. cooldown_interval : float = 0.0 The cool down interval between two measurements in seconds. enable_cpu_cache_flush: bool = False Whether to flush cache on CPU between repeated measurements. Flushing cache can make the measured latency of one operator closer to its actual latency during end-to-end inference. To make this option effective, the argument `number` should also be set to 1. This is only has effect on CPU task. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during program measuring. device: int = 0 Which device to run on if multiple are available. Returns ------- res : List[MeasureResult] The measure results of these MeasureInputs. """ measure_results = [] assert len(inputs) == len(build_results), "Measure input size should be equal to build results" worker = PopenWorker() for inp, build_res in zip(inputs, build_results): if build_res.error_no != 0: res = ( (MAX_FLOAT,), build_res.error_no, build_res.error_msg, build_res.time_cost, time.time(), ) else: args = prepare_runner_args(inp, build_res) res = call_func_with_timeout( worker, timeout, _timed_eval_func, args=( inp.serialize(), build_res, args, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, verbose, device, ), ) if isinstance(res, TimeoutError): if verbose >= 1: print("*T", end="", flush=True) # Run timeout res = ( (MAX_FLOAT,), MeasureErrorNo.RUN_TIMEOUT, None, build_res.time_cost + timeout, time.time(), ) elif isinstance(res, Exception): if verbose >= 1: print("*E", end="", flush=True) # Run error res = ( (MAX_FLOAT,), MeasureErrorNo.RUNTIME_DEVICE, str(res), build_res.time_cost + timeout, time.time(), ) measure_results.append(MeasureResult(*res)) if verbose >= 1: print("", flush=True) return measure_results def _rpc_run( inp_serialized, build_res, args, key, host, port, priority, timeout, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, verbose, device, ): inp = MeasureInput.deserialize(inp_serialized) tic = time.time() error_no = 0 error_msg = None try: # upload built module remote = request_remote(key, host, port, priority, timeout) remote.upload(build_res.filename) func = remote.load_module(os.path.split(build_res.filename)[1]) dev = remote.device(str(inp.task.target), device) # Limitation: # We can not get PackFunction directly in the remote mode as it is wrapped # under the std::function. We could lift the restriction later once we fold # the PackedFunc as an object. Currently, we pass function name to work # around it. f_prepare = "cache_flush_cpu_non_first_arg" if enable_cpu_cache_flush else "" time_f = func.time_evaluator( func.entry_name, dev, number=number, repeat=repeat, min_repeat_ms=min_repeat_ms, f_preproc=f_prepare, ) # pylint: disable=broad-except except Exception: costs = (MAX_FLOAT,) error_no = MeasureErrorNo.COMPILE_DEVICE error_msg = make_traceback_info() if error_no == 0: try: stream = dev.create_raw_stream() dev.set_raw_stream(stream) random_fill = remote.get_function("tvm.contrib.random.random_fill") assert ( random_fill ), "Please make sure USE_RANDOM is ON in the config.cmake on the remote devices" assert len(args) == len(build_res.args) loc_args = [] # pylint: disable=consider-using-enumerate for idx in range(len(args)): if args[idx] is None: build_res_arg = build_res.args[idx] empty_array = ndarray.empty( get_const_tuple(build_res_arg.shape), build_res_arg.dtype, dev ) random_fill(empty_array) loc_args.append(empty_array) else: loc_args.append(ndarray.array(args[idx], dev)) dev.sync() # First run for check that the kernel is correct func.entry_func(*loc_args) dev.sync() costs = time_f(*loc_args).results # clean up remote files remote.remove(build_res.filename) remote.remove(os.path.splitext(build_res.filename)[0] + ".so") remote.remove("") dev.free_raw_stream(stream) # pylint: disable=broad-except except Exception: dev.free_raw_stream(stream) costs = (MAX_FLOAT,) error_no = MeasureErrorNo.RUNTIME_DEVICE error_msg = make_traceback_info() shutil.rmtree(os.path.dirname(build_res.filename)) toc = time.time() time.sleep(cooldown_interval) if verbose >= 1: if error_no == MeasureErrorNo.NO_ERROR: print("*", end="") else: print("*E", end="") # Run error return costs, error_no, error_msg, toc - tic + build_res.time_cost, toc def _rpc_run_worker(args): """Function to be ran in the RPCRunner thread pool. Parameters ---------- args : Tuple[MeasureInput, BuildResult, ...] Single input and build result plus the rest of the arguments to `rpc_runner_run`. Returns ------- res : MeasureResult The measure result of this Runner thread. """ _, build_res, _, _, _, _, _, timeout, _, _, _, _, _, verbose, _ = args if build_res.error_no != MeasureErrorNo.NO_ERROR: return ( (MAX_FLOAT,), build_res.error_no, build_res.error_msg, build_res.time_cost, time.time(), ) try: res = _rpc_run(*args) # pylint: disable=broad-except except Exception: if verbose >= 1: print("*E", end="") # Run error res = ( (MAX_FLOAT,), MeasureErrorNo.RUNTIME_DEVICE, make_traceback_info(), build_res.time_cost + timeout, time.time(), ) return res @tvm._ffi.register_func("auto_scheduler.rpc_runner.run") def rpc_runner_run( inputs, build_results, key, host, port, priority=1, n_parallel=1, timeout=10, number=3, repeat=1, min_repeat_ms=0, cooldown_interval=0.0, enable_cpu_cache_flush=False, verbose=1, device=0, ): """Run function of RPCRunner to test the performance of the input BuildResults. Parameters ---------- inputs : List[MeasureInput] The MeasureInputs to be measured. build_results : List[BuildResult] The BuildResults to be measured. key : str The key of the device registered in the RPC tracker. host : str The host address of the RPC Tracker. port : int The port of RPC Tracker. priority : int = 1 The priority of this run request, larger is more prior. n_parallel : int = 1 The number of tasks run in parallel. timeout : int = 10 The timeout limit (in second) for each run. This is used in a wrapper of the multiprocessing.Process.join(). number : int = 3 The number of times to run the generated code for taking average. We call these runs as one `repeat` of measurement. repeat : int = 1 The number of times to repeat the measurement. In total, the generated code will be run (1 + number x repeat) times, where the first "1" is warm up and will be discarded. The returned result contains `repeat` costs, each of which is an average of `number` costs. min_repeat_ms : int = 0 The minimum duration of one `repeat` in milliseconds. By default, one `repeat` contains `number` runs. If this parameter is set, the parameters `number` will be dynamically adjusted to meet the minimum duration requirement of one `repeat`. i.e., When the run time of one `repeat` falls below this time, the `number` parameter will be automatically increased. cooldown_interval : float = 0.0 The cool down interval between two measurements in seconds. enable_cpu_cache_flush: bool = False Whether to flush cache on CPU between repeated measurements. Flushing cache can make the measured latency of one operator closer to its actual latency during end-to-end inference. To make this option effective, the argument `number` should also be set to 1. This is only has effect on CPU task. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during program measuring. device: int = 0 Which device to run on if multiple are available. Returns ------- res : List[MeasureResult] The measure results of these MeasureInputs. """ assert len(inputs) == len(build_results), "Measure input size should be equal to build results" # This pool is not doing computationally intensive work, so we can use threads executor = PopenPoolExecutor(n_parallel) tuple_res = executor.map_with_error_catching( _rpc_run_worker, [ ( inp.serialize(), build_res, prepare_runner_args(inp, build_res), key, host, port, priority, timeout, number, repeat, min_repeat_ms, cooldown_interval, enable_cpu_cache_flush, verbose, device, ) for inp, build_res in zip(inputs, build_results) ], ) results = [] for i, res in enumerate(tuple_res): if res.status == StatusKind.COMPLETE: results.append(MeasureResult(*res.value)) else: assert res.status == StatusKind.TIMEOUT if verbose >= 1: print("*T", end="") # Run timeout build_res = build_results[i] results.append( MeasureResult( (MAX_FLOAT,), MeasureErrorNo.RUN_TIMEOUT, None, build_res.time_cost + timeout, time.time(), ) ) if verbose >= 1: print("") return results
45,878
33.366292
100
py
tvm
tvm-main/python/tvm/auto_scheduler/search_task.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. """ The definiton of SearchTask """ import json import os import logging import numpy as np import tvm._ffi from tvm.runtime import Object, ndarray from tvm.driver.build_module import build from tvm.target import Target from .measure import LocalBuilder, LocalRunner from .measure_record import load_best_record from .workload_registry import make_workload_key from .compute_dag import ComputeDAG, LayoutRewriteOption from .cost_model import XGBModel from .search_policy import SketchPolicy from .workload_registry import WORKLOAD_FUNC_REGISTRY, register_workload_tensors from . import _ffi_api # pylint: disable=invalid-name logger = logging.getLogger("auto_scheduler") @tvm._ffi.register_object("auto_scheduler.HardwareParams") class HardwareParams(Object): """The parameters of target hardware used to guide the search policy. When a parameter isn't provided, it will instead use the current machine's default value if target is specified. TODO(jcf94): This is considered to be merged with the new Target specification: https://discuss.tvm.apache.org/t/rfc-tvm-target-specification/6844 Parameters ---------- num_cores : int, optional The number of device cores. vector_unit_bytes : int, optional The width of vector units in bytes. cache_line_bytes : int, optional The size of cache line in bytes. max_shared_memory_per_block : int, optional The max shared memory per block in bytes. max_local_memory_per_block : int, optional The max local memory per block in bytes. max_threads_per_block : int, optional The max number of threads per block. max_vthread_extent : int, optional The max vthread extent. warp_size : int, optional The thread numbers of a warp. target : str or Target, optional The compilation target. Used to determine default values if provided. target_host : str or Target, optional The compilation target host. Used to determine default values if provided. """ def __init__( self, num_cores=None, vector_unit_bytes=None, cache_line_bytes=None, max_shared_memory_per_block=None, max_local_memory_per_block=None, max_threads_per_block=None, max_vthread_extent=None, warp_size=None, target=None, target_host=None, ): # If target is provided, get the default paramters for this machine. if target is not None: if isinstance(target, str): target = tvm.target.Target(target) if isinstance(target_host, str): target_host = tvm.target.Target(target_host) default_params = _ffi_api.GetDefaultHardwareParams(target, target_host) if num_cores is None: num_cores = default_params.num_cores if vector_unit_bytes is None: vector_unit_bytes = default_params.vector_unit_bytes if cache_line_bytes is None: cache_line_bytes = default_params.cache_line_bytes if max_shared_memory_per_block is None: max_shared_memory_per_block = default_params.max_shared_memory_per_block if max_local_memory_per_block is None: max_local_memory_per_block = default_params.max_local_memory_per_block if max_threads_per_block is None: max_threads_per_block = default_params.max_threads_per_block if max_vthread_extent is None: max_vthread_extent = default_params.max_vthread_extent if warp_size is None: warp_size = default_params.warp_size self.__init_handle_by_constructor__( _ffi_api.HardwareParams, num_cores, vector_unit_bytes, cache_line_bytes, max_shared_memory_per_block, max_local_memory_per_block, max_threads_per_block, max_vthread_extent, warp_size, ) def __str__(self): """Pretty printing for hardware parameter configuration.""" format_str = ( "HardwareParams:\n" f" num_cores: {self.num_cores}\n" f" vector_unit_bytes: {self.vector_unit_bytes}\n" f" cache_line_bytes: {self.cache_line_bytes}\n" f" max_shared_memory_per_block: {self.max_shared_memory_per_block}\n" f" max_local_memory_per_block: {self.max_local_memory_per_block}\n" f" max_threads_per_block: {self.max_threads_per_block}\n" f" max_vthread_extent: {self.max_vthread_extent}\n" f" warp_size: {self.warp_size}\n" ) return format_str @tvm._ffi.register_object("auto_scheduler.TuningOptions") class TuningOptions(Object): """This controls the options of performance tuning. Parameters ---------- num_measure_trials: int = 0 The number of measurement trials. The search policy measures `num_measure_trials` schedules in total and returns the best one among them. With `num_measure_trials` == 0, the policy will do the schedule search but won't involve measurement. This can be used to get a runnable schedule quickly without auto-tuning. early_stopping: Optional[int] Stop the tuning early if getting no improvement after n measurements. num_measures_per_round: int = 64 The number of schedules to be measured at each search round. The whole schedule search process will try a total number of `num_measure_trials` in several rounds. verbose: int = 1 Verbosity level. 0 for silent, 1 to output information during schedule search. builder: Union[ProgramBuilder, str] = 'local' ProgramBuilder which builds the program. runner: Union[ProgramRunner, str] = 'local' ProgramRunner which runs the program and measures time costs. measure_callbacks: Optional[List[MeasureCallback]] Callback functions called after each measurement. Candidates: - auto_scheduler.RecordToFile """ def __init__( self, num_measure_trials=0, early_stopping=None, num_measures_per_round=64, verbose=1, builder="local", runner="local", measure_callbacks=None, ): if isinstance(builder, str): if builder == "local": builder = LocalBuilder() else: raise ValueError("Invalid builder: " + builder) elif not isinstance(builder, tvm.auto_scheduler.measure.ProgramBuilder): raise ValueError( "Invalid builder: " + builder + " . TuningOptions expects a ProgramBuilder or string." ) if isinstance(runner, str): if runner == "local": runner = LocalRunner() else: raise ValueError("Invalid runner: " + runner) elif not isinstance(runner, tvm.auto_scheduler.measure.ProgramRunner): raise ValueError( "Invalid runner: " + runner + " . TuningOptions expects a ProgramRunner or string." ) self.__init_handle_by_constructor__( _ffi_api.TuningOptions, num_measure_trials, early_stopping or -1, num_measures_per_round, verbose, builder, runner, measure_callbacks, ) # The map stores special registered buffer for measurement. # This can be used for sparse workloads when we cannot use random tensors for measurment. # { # "workload_key_0": { # "task_input_0": Tensor(...), # "task_input_1": Tensor(...) # }, # "workload_key_1": { # "task_input_2": Tensor(...), # "task_input_3": Tensor(...) # }, # ... # } TASK_INPUT_BUFFER_TABLE = {} def _save_buffer_to_file(buffer_name, buffer_data): """Save the current Tensor buffer to a numpy file. File name will be: {buffer_name}.{buffer_shape}_{buffer_data_type}.npy """ np_data = buffer_data.numpy() buffer_name += "." for i in np_data.shape: buffer_name += f"{i}_" buffer_name += f"{np_data.dtype}.npy" np_data.tofile(buffer_name, " ") def _try_load_buffer_from_file(buffer_name): """Try to load buffer from a numpy file, if not found, return None. File name has a same format as `_save_buffer_to_file`. """ filelist = os.listdir() for file in filelist: if file.startswith(buffer_name + "."): meta_info = file.split(".")[-2].split("_") shape = [int(i) for i in meta_info[:-1]] dtype = meta_info[-1] buffer_data = np.fromfile(file, dtype=dtype, sep=" ") buffer_data = buffer_data.reshape(shape) return ndarray.array(buffer_data) return None def register_task_input_buffer( workload_key, input_name, input_data, overwrite=False, save_to_file=False ): """Register special buffer for measurement. Parameters ---------- workload_key : str The workload key of the SearchTask. input_name : str The name of input buffer. input_data : tvm.nd.NDArray The input Tensor data. overwrite : bool = False Whether to overwrite the data if a name has already registered. save_to_file : bool = False Whether to save the data to a local file as well. This can be reused to resume the last tuning process. Returns ------- tvm.nd.NDArray The actual registered Tensor data of this input_name. With `overwrite` set to False, will return the original one if the name has already registered before. """ global TASK_INPUT_BUFFER_TABLE if workload_key not in TASK_INPUT_BUFFER_TABLE: TASK_INPUT_BUFFER_TABLE[workload_key] = {} input_table = TASK_INPUT_BUFFER_TABLE[workload_key] if not overwrite: if input_name not in input_table.keys(): # Try to load buffer data from local file tensor_from_file = _try_load_buffer_from_file(input_name) if tensor_from_file: input_table[input_name] = tensor_from_file elif input_name in input_table.keys(): raise RuntimeError( "Tensor %s exists in TASK_INPUT_BUFFER_TABLE, %s" % (input_name, "set overwrite to True or this Tensor will not be registered") ) input_table[input_name] = input_data if save_to_file: _save_buffer_to_file(input_name, input_data) return input_data def get_task_input_buffer(workload_key, input_name): """Get special buffer for measurement. The buffers are registered by `register_task_input_buffer`. Parameters ---------- workload_key : str The workload key of the SearchTask. input_name : str The name of input buffer. Returns ------- tvm.nd.NDArray The registered input buffer. """ global TASK_INPUT_BUFFER_TABLE if workload_key not in TASK_INPUT_BUFFER_TABLE: TASK_INPUT_BUFFER_TABLE[workload_key] = {} input_table = TASK_INPUT_BUFFER_TABLE[workload_key] if input_name not in input_table: # Try to load buffer data from local file tensor_from_file = _try_load_buffer_from_file(input_name) if tensor_from_file: input_table[input_name] = tensor_from_file # Then check for the default table, the input names extracted from a relay model will be # stored here for we're not able to get the workload_key at that time if input_name not in input_table: input_table = TASK_INPUT_BUFFER_TABLE["default"] if input_name in input_table: return input_table[input_name] raise ValueError( f"{input_name} not found in TASK_INPUT_BUFFER_TABLE, " f"should provide with `SearchTask(..., task_inputs={{...}})`" ) @tvm._ffi.register_object("auto_scheduler.SearchTask") class SearchTask(Object): """The computation information and hardware parameters for a schedule search task. Parameters ---------- func : Union[Function, str] The function that returns the compute declaration Tensors. Can be the a function or the function name. args : Union[Tuple[Any, ...], List[Any]] The args of the function. compute_dag : ComputeDAG The ComputeDAG for the corresponding compute declaration. workload_key : str The workload key for the corresponding compute declaration. target : any target-like object, see Target.canon_target The target device of this search task. target_host : None or any target-like object, see Target.canon_target The target host device of this search task. hardware_params : Optional[HardwareParams] Hardware parameters used in this search task. layout_rewrite_option : Optional[LayoutRewriteOption] The layout rewrite option used for measuring programs. If None, the default value will be set depending on the specified target. Auto_scheduler will find a better schedule for the specified layout rewrite option. The NO_REWRITE and INSERT_TRANSFORM_STAGE are expected to be used when tuning a standalone op, and the REWRITE_FOR_PRE_TRANSFORMED is expected to be used when tuning ops inside a network. task_inputs : Union[Dict[str, tvm.nd.NDArray], List[str]] A dict maps the input names to input tensors or a list of input names. Some special Tensor used as inputs in program measuring. Usually we do not need to care about it, but for special workloads like Sparse computation the Sparse Tensor input are meaningful that we cannot use random input directly. task_inputs_overwrite : bool = False Whether to overwrite the data if a name has already in the global table. task_inputs_save_to_file : bool = False Whether to save the data to a local file as well. This can be reused to resume the last tuning process. desc: str = "" The description string of this task. Examples -------- .. code-block:: python # We support two ways to create a search task # Way 1: create a task by a workload generation function. # The `workload_func` is a function decorated by @auto_scheduler.register_workload task = SearchTask(func=workload_func, args=args, target=target) # Way 2: create a task by a workload_key. # The `workload_key` is a string, which can be either a hash key or a json-serialized # tuple(func, args). task = SearchTask(workload_key=workload_key, target=target) """ def __init__( self, func=None, args=None, compute_dag=None, workload_key=None, target=None, target_host=None, hardware_params=None, layout_rewrite_option=None, task_inputs=None, task_inputs_overwrite=False, task_inputs_save_to_file=False, desc="", ): assert ( func is not None or workload_key is not None ), "Either a workload generation function or a workload key should be provided" if func is not None: workload_key = make_workload_key(func, args) if compute_dag is None: compute_dag = ComputeDAG(workload_key) assert target is not None, "Must specify a target." target, target_host = Target.canon_target_and_host(target, target_host) if layout_rewrite_option is None: layout_rewrite_option = LayoutRewriteOption.get_target_default(target) task_input_names = [] if isinstance(task_inputs, list): task_input_names = task_inputs elif isinstance(task_inputs, dict): for input_name in task_inputs: register_task_input_buffer( workload_key, input_name, task_inputs[input_name], task_inputs_overwrite, task_inputs_save_to_file, ) task_input_names.append(input_name) elif task_inputs is not None: raise ValueError("task_inputs should be a dict or a list.") self.__init_handle_by_constructor__( _ffi_api.SearchTask, compute_dag, workload_key, target, target_host, hardware_params, layout_rewrite_option, task_input_names, desc, ) def tune(self, tuning_options, search_policy=None, adaptive_training=False): """Run auto scheduling search for a task Parameters ---------- tuning_options : TuningOptions Tuning and measurement options. search_policy : Optional[SearchPolicy] The search policy to be used for schedule search. """ if search_policy is None: cost_model = XGBModel(adaptive_training=adaptive_training) search_policy = SketchPolicy(self, cost_model) _ffi_api.AutoSchedule(search_policy, tuning_options) def apply_best(self, log_file, include_compatible=False, layout_rewrite_option=None): """Apply the history best from a log file and return the schedule. Parameters ---------- log_file : str The name of the log file. include_compatible: bool When set to True, all compatible records in the log file will be considered. layout_rewrite_option : Optional[LayoutRewriteOption] The layout rewrite option. Returns ------- A `te.Schedule` and the a list of `te.Tensor` to be used in `tvm.lower` or `tvm.build`. """ inp, _ = load_best_record( log_file, self.workload_key, include_compatible=include_compatible ) if inp is None: raise RuntimeError( f"Cannot find any valid schedule for {self.workload_key} in file {log_file}" ) sch, args = self.compute_dag.apply_steps_from_state( inp.state, layout_rewrite_option or self.layout_rewrite_option ) return sch, args def print_best(self, log_file, print_mode="schedule"): """Print the best schedule as python schedule API code or CUDA source code. Parameters ---------- log_file : str The name of the log file print_mode: str if "schedule", print the best schedule as python schedule API code. if "cuda", print the best schedule as CUDA source code. Returns ------- code: str The best schedule code in python API or CUDA source code """ inp, _ = load_best_record(log_file, self.workload_key) if inp is None: raise RuntimeError( f"Cannot find any valid schedule for {self.workload_key} in file {log_file}" ) if print_mode == "schedule": return self.compute_dag.print_python_code_from_state(inp.state) if print_mode == "cuda": assert self.target.kind.name == "cuda" sch, args = self.compute_dag.apply_steps_from_state(inp.state) func = build(sch, args, "cuda") return func.imported_modules[0].get_source() raise ValueError(f"Invalid print_mode: {print_mode}") def __getstate__(self): self.target, self.target_host = Target.canon_target_and_host(self.target, self.target_host) return { "compute_dag": self.compute_dag, "workload_key": self.workload_key, "target": self.target, "target_host": self.target_host, "hardware_params": self.hardware_params, "layout_rewrite_option": self.layout_rewrite_option, "task_input_names": self.task_input_names, "desc": self.desc, } def __setstate__(self, state): # Register the workload if needed try: workload = json.loads(state["workload_key"]) except Exception: # pylint: disable=broad-except raise RuntimeError(f"Invalid workload key {state['workload_key']}") # workload[0] is either the compute function name or the ComputeDAG hash. # The compute functions are already registered when importing TVM, so here # we only register the ComputeDAG workloads. If the same workload has # already been registered, the later registration overrides the prvious one. if workload[0] not in WORKLOAD_FUNC_REGISTRY: register_workload_tensors(state["workload_key"], state["compute_dag"].tensors) state["target"], state["target_host"] = Target.canon_target_and_host( state["target"], state["target_host"] ) self.__init_handle_by_constructor__( _ffi_api.SearchTask, state["compute_dag"], state["workload_key"], state["target"], state["target"].host, state["hardware_params"], state["layout_rewrite_option"], state["task_input_names"], state["desc"], ) def create_task(func, args, target, target_host=None, hardware_params=None): """THIS API IS DEPRECATED. Create a search task. Parameters ---------- func : Union[Function, str] The function that returns the compute declaration Tensors. Can be the a function or the function name. args : Union[Tuple[Any, ...], List[Any]] The args of the function. target : Union[tvm.target.Target, str] The target device of this search task. target_host : Optional[Union[tvm.target.Target, str]] The target host device of this search task. hardware_params : Optional[HardwareParams] Hardware parameters used in this search task. Returns ------- SearchTask: the created task """ raise ValueError( 'The API "auto_scheduler.create_task" is deprecated.' "See https://github.com/apache/tvm/pull/7028 for the upgrade guide" ) def auto_schedule(task, search_policy=None, tuning_options=TuningOptions()): """THIS API IS DEPRECATED. Run auto scheduling search for a task. Parameters ---------- task : SearchTask The SearchTask for the computation declaration. search_policy : Optional[SearchPolicy] The search policy to be used for schedule search. tuning_options : Optional[TuningOptions] Tuning and measurement options. Returns ------- A `te.Schedule` and the a list of `te.Tensor` to be used in `tvm.lower` or `tvm.build`. """ raise ValueError( 'The API "auto_scheduler.create_task" is deprecated.' "See https://github.com/apache/tvm/pull/7028 for the upgrade guide." )
23,856
35.703077
100
py
tvm
tvm-main/python/tvm/auto_scheduler/__init__.py
# Licensed to the Apache Software Foundation (ASF) under one # or more contributor license agreements. See the NOTICE file # distributed with this work for additional information # regarding copyright ownership. The ASF licenses this file # to you under the Apache License, Version 2.0 (the # "License"); you may not use this file except in compliance # with the License. You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, # software distributed under the License is distributed on an # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY # KIND, either express or implied. See the License for the # specific language governing permissions and limitations # under the License. # pylint: disable=unused-import, redefined-builtin """ Namespace for TVM Auto-scheduler. """ from . import ( compute_dag, dispatcher, feature, loop_state, measure, measure_record, relay_integration, search_policy, search_task, task_scheduler, utils, workload_registry, ) # Shortcut from .compute_dag import ( ComputeDAG, LayoutRewriteOption, get_shape_from_rewritten_layout, ) from .cost_model import RandomModel, XGBModel from .dispatcher import ApplyHistoryBest, ApplyHistoryBestOrSample, DispatchContext from .measure import ( LocalBuilder, LocalRPCMeasureContext, LocalRunner, MeasureInput, MeasureResult, RPCRunner, register_task_input_check_func, ) from .measure_record import ( RecordReader, RecordToFile, load_best_record, load_records, save_records, ) from .relay_integration import ( extract_tasks, is_auto_scheduler_enabled, remove_index_check, rewrite_compute_body, rewrite_tensor_shape, ) from .search_policy import ( EmptyPolicy, PreloadCustomSketchRule, PreloadMeasuredStates, SketchPolicy, ) from .search_task import ( HardwareParams, SearchTask, TuningOptions, auto_schedule, create_task, ) from .task_scheduler import TaskScheduler from .workload_registry import make_workload_key, register_workload
2,169
25.790123
83
py