text stringlengths 0 1.15k |
|---|
Academic teams will do that, too. We've got some collaborators who built parallel programming models who've gone and spun off companies for things like that. |
There are a whole lot of ways to get funded. Navigating all that I think is fairly difficult, and then one observation that I guess I would say I have about the whole process is that there is a lot of research funding in our area for cutting edge things, but as far as actually maintaining the software, that is not cons... |
I think doing that is often a lot of work... You need to socialize your thing, you need to convince someone who controls programmatic funding that they need this thing. Often times, that's getting it into one of our simulation codes, making them rely on it; that's something that would catch people's interest. |
Other things would be making it a critical product for the compute center, like if we're actually using it to deploy clusters and we need it, and it makes us more efficient - that's another way that you can get programmatic funding. But in general, it depends on what the software is and what mission it supports how you... |
There's also a distinction between hard money and soft money. Are you guys familiar with that from Academic research? |
**Nadia Eghbal:** No. |
**Todd Gamblin:** Hard money would be like the programmatic money, where it's ongoing, it doesn't necessarily go away... It's either overhead for the organization, or it's part of how Livermore is supported over time. Soft money is stuff that you have to apply for, and it tends to have a short lifetime. If you get a re... |
Usually, the exit plan for a lot of projects here is get programmatic funding for the thing, but that doesn't always happen. It's not an easy task to get a project funded, I would say, and to get it to grown. I think that's pretty similar to elsewhere. |
**Nadia Eghbal:** Nobody cares about maintenance, huh? |
**Todd Gamblin:** Right, and I think that's actually -- I don't think people think about maintenance as much as they should. I think people think that once you do the research, that it's magic, or something, and the software continues to work because you've already done the development for it... But there's a ton of ma... |
\[48:07\] Depending on the part of the organization it is, the compute center here is very conscious of the maintenance costs. I don't necessarily think that the people who are running research programs are always aware of that, or thinking about how that thing might take off after the research program. I mean, that's ... |
Tech Transfer is something that we care a lot about. We're supposed to make products viable for industry, and people like it when you do that... But on the software side, I think that -- I think one thing is that the labs... Like, we're fundamentally a science organization, right? We have a science mission, and we have... |
For example, for our simulation codes I think there was this study done, and we found that the number of engineers per lines of code that they had to support and maintain on our teams was a lot -- we had a lot more lines of code per engineers than industry teams do. That was interesting to me, just from a "what do we t... |
**Mikeal Rogers:** I feel like we're covering some of the instability in the funding and we're kind of focused on that, and what it doesn't do... But I'm just trying to compare this to other funding sources that we've talked about. If you start a startup, you're gonna have to go out and beg for money every two years... |
**Todd Gamblin:** Yes. |
**Mikeal Rogers:** If you do grant funding, you're gonna spend a year to get a year of funding, and then have to beg for money again, year on year on year for grants. I think compared to those funding sources, it's actually relatively stable. |
**Todd Gamblin:** It is. Programmatic funding is definitely stable compared to those, and I think that's a good thing, right? I think you've talked on this show in the past about software as a public good, right? Maybe the infrastructure is something that should be funded by some part of government for maintenance. I t... |
On our side, the things that I'm pushing for are -- we've gotta think about things that could have a broader impact, we've gotta think about things that we could build communities around, because then we could get both contributors for our projects, and also the investment would pay off for us because we'd be supportin... |
That's actually something that I've tried to point out with Spack - I have this chart that shows "Here's the amount of programmatic funding that we have for Spack", it's like two engineers, or something like that, and "Here's all the contributions that we've gotten." Over the course of a year, the number of packages de... |
**Nadia Eghbal:** I'm wondering, for getting contributions specifically... That must be influenced by academic cycles as well, right? So if you have people that are contributing while they're doing their post-doc, and afterwards they stop - is that a thing? |
**Todd Gamblin:** \[52:04\] Yeah, that's a really good point. I think that might be another aspect of the research funding structure that maybe doesn't work so well as the long-term sustainability strategy. In research you're really encouraged to be a PI (principle investigator) and to start new things, so to advance y... |
I think some projects have managed to make it do that, projects that are widely used... I mentioned the ROSE compiler here. They've published tons of papers, and they always have new collaboration going on some piece of their compiler infrastructure. But there's only certain products that can do that. |
We have a paper on Spack, we submitted it to the State-of-the-Practice Track at Supercomputing, which is the big conference in my area, and I think that got us a lot of publicity, but how we would publish more papers, or continue to publish papers about it is up in the air, so we have to rely on the programmatic fundin... |
**Nadia Eghbal:** Can you talk a little bit about Exascale? Because I know you mentioned that that's a project where the focus has been on actually developing the software and not just on writing papers or getting funding that way, so how did that happen? |
**Todd Gamblin:** The Exascale computing project is a major collaboration between the six national laboratories. So it's Livermore, Los Alamos, Sandia, Argonne, Oak Ridge and Lawrence Berkeley Lab. PNNL is in there as well - that's Pacific Northwest National Lab... And I think some other labs are collaborating. But it'... |
So there are plans to build large Exascale machines, and this project is to design the software stack. That's at exascaleproject.org. That was sort of motivated by an immediate need to go and actually build the software stack so that when the Exascale machine gets here - which is something that we're planning to do - t... |
There's 15 or so simulation codes in the Exascale project, everything from like nuclear fusion to the climate community, to molecular dynamics... There's lots of useful science there that could be used and could have an impact on industries, like for making cars better, making more clean energy in different areas... Bu... |
I'm involved in it in two ways - one is with Spack. I would like Spack to be the package manager that's used for the Exascale project, so that that's how we deploy things on supercomputers, that's how we built things and that's how we make it easy for, say, someone in the industry to pick up one of these codes and use ... |
I'm also the lead on a software productivity project - I'm the Livermore lead on a software productivity project (Mike Heroux and Lois Curfman-McInnes from Sandia and Argonne are the main leads on this). But there's actually like an effort within this project, because it has to deliver software, to make the developers ... |
That involves things like putting training out there, familiarizing scientists and some of these computational scientists who develop the applications with how to build communities and how to write documentation, how to use source control well, how to have a release cycle, things like that. |
\[56:12\] I think it's kind of new territory in a lot of ways, because I don't think we've had a coordinated effort to build a software stack quite as large, so one of the things we've been talking about lately is "How do we coordinate releases of 15 applications in 80 different software projects?" That's not easy. And... |
**Nadia Eghbal:** And just to clarify, that is an open source software stack... |
**Todd Gamblin:** So not all of it is open source. For example, some of our weapons simulations are included in the Exascale project, because they need to be able to run on Exascale machines, but those are not open source. Sorry. |
**Nadia Eghbal:** Phew! |
**Todd Gamblin:** Yeah, but in large part -- a lot of the science parts of it and a lot of the math libraries and the computer science infrastructure are open source; this is stuff that people could build on. I think it would be really awesome if we could build simulation frameworks and things that someone could come a... |
Our code teams are sort of starting to think about that. We recently had an effort internally to look at how our simulations are structured, and I think they found that something like 40%-50% of the code is pure computer science - no physics, nothing sensitive - and that part could be factored out as like a general too... |
**Nadia Eghbal:** So how do you get industry contributors involved in situations like that? Do they come to the same sort of conferences that you do? Do you have to reach out to them individually? |
**Todd Gamblin:** Yeah, that's a problem. They don't always come to the same sorts of conferences. Industry HPC has always been -- it's something that people shoot for, but not a lot of companies really get into it. There are companies like Exxon that have HPC clusters and they kind of do their own specific simulations... |
But one of the goals of this project would be to sort of expand the middle tier of HPC. It's always been sort of elusive. We'd like to have smaller companies able to use these kinds of resources, but I think the complexity of getting into supercomputing has been so high that they haven't necessarily jumped on board. Bu... |
I think Procter & Gamble - I'm blanking on the guy's name, but he came out here and gave a talk about all the different ways that they use HPC. You've heard of Procter & Gamble I guess, but they have lots of companies under them that they run, so he talked about how they'd used simulations for everything from making th... |
The other problem with getting industry contributors I think is that in many cases for something as complex as like a piece of simulation software, the industry folks really want someone to support it, they want someone to call and to say "We're having problems deploying this. What do we do?" I think for that we would ... |
**Nadia Eghbal:** \[01:00:06.16\] Well, we're gonna close out on that note... Do you have any final thoughts to share about lessons learned from your experiences open sourcing in a fairly (I'd say) comparably difficult context? |
**Todd Gamblin:** One of the things that I've learned from Spack I guess is that you really have to think about the broader context for the thing that you're building, and that means giving up some degree of control. I think there are a lot of projects where they've grown up in one lab and they've served that one team,... |
For Spack we've tried to be really open about contributions; maybe not necessarily implementing things for people, but helping them to implement things and thinking outside of our own use case. That's one of the ways that I think we've gotten contributors and that's how I think we've been able to grow our community. |
**Mikeal Rogers:** I think that's a general open source problem... A lot of people who start projects don't wanna give up control, but they want people to work on it. |
**Todd Gamblin:** Yeah, I totally agree. And teaching people to do that... Actually, that's one thing I could say - a lot of our teams are scared of what will happen if they put something out there as open source and they start to get a lot of pull requests. They're like, "What do I do? I'm gonna have to support all th... |
I don't have a silver bullet for "How do we maintain this feature you contributed after you contributed it?" I think that's the harder problem and you have to get creative. |
**Mikeal Rogers:** Definitely. Thanks for coming on. |
**Todd Gamblin:** Sure, thanks for having me. It's been great! |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.