Index individual PyC kernel repositories
Browse filesLink each documented lane to its dedicated Kernel repository.
README.md
CHANGED
|
@@ -45,3 +45,27 @@ The source files are organized by prototype family; several documented lanes
|
|
| 45 |
share one parameterized `.cu` implementation. See the companion
|
| 46 |
[performance Space](https://huggingface.co/spaces/AethronPhantom/pyc-kernel-lab)
|
| 47 |
for the optimization story and selected measurements.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
share one parameterized `.cu` implementation. See the companion
|
| 46 |
[performance Space](https://huggingface.co/spaces/AethronPhantom/pyc-kernel-lab)
|
| 47 |
for the optimization story and selected measurements.
|
| 48 |
+
|
| 49 |
+
## Individual Kernel Repositories
|
| 50 |
+
|
| 51 |
+
Each documented lane is also available as its own Hugging Face Kernel repository:
|
| 52 |
+
|
| 53 |
+
- [matrix_mult](https://huggingface.co/kernels/AethronPhantom/matrix_mult)
|
| 54 |
+
- [tokenizer_kernel](https://huggingface.co/kernels/AethronPhantom/tokenizer_kernel)
|
| 55 |
+
- [ada_gemm](https://huggingface.co/kernels/AethronPhantom/ada_gemm)
|
| 56 |
+
- [ada_gemm_k64_warp32_async](https://huggingface.co/kernels/AethronPhantom/ada_gemm_k64_warp32_async)
|
| 57 |
+
- [ada_tensor_core_fp16](https://huggingface.co/kernels/AethronPhantom/ada_tensor_core_fp16)
|
| 58 |
+
- [ada_tensor_core_bf16](https://huggingface.co/kernels/AethronPhantom/ada_tensor_core_bf16)
|
| 59 |
+
- [hopper_tensor_core_fp16](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_fp16)
|
| 60 |
+
- [hopper_tensor_core_bf16](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16)
|
| 61 |
+
- [hopper_tensor_core_bf16_perf](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_perf)
|
| 62 |
+
- [hopper_tensor_core_bf16_warp2n](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_warp2n)
|
| 63 |
+
- [hopper_tensor_core_bf16_k32](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_k32)
|
| 64 |
+
- [hopper_tensor_core_bf16_warp2n_k32](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_warp2n_k32)
|
| 65 |
+
- [hopper_cublaslt_bf16](https://huggingface.co/kernels/AethronPhantom/hopper_cublaslt_bf16)
|
| 66 |
+
- [hopper_cublaslt_bf16_check](https://huggingface.co/kernels/AethronPhantom/hopper_cublaslt_bf16_check)
|
| 67 |
+
- [hopper_tensor_core_bf16_async](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_async)
|
| 68 |
+
- [hopper_tensor_core_bf16_async_wide](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_async_wide)
|
| 69 |
+
- [hopper_tensor_core_bf16_async_wide_k64](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_async_wide_k64)
|
| 70 |
+
- [hopper_tensor_core_bf16_async_square](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_async_square)
|
| 71 |
+
- [hopper_tensor_core_bf16_async_square_k64](https://huggingface.co/kernels/AethronPhantom/hopper_tensor_core_bf16_async_square_k64)
|