| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: Mandatory "Request for Project to be Ported to GPU" Thread |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 133
|
|
| Author |
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1326 Status: Offline Project Badges: 
|
If it happens I'd like to vote for nVidia OpenCL 1.2 They will get there faster with CUDA, which is what Quarantine@Home is using for their Autodock project. https://quarantine.infino.me/ As far as I am aware the GPU version is only supported by Linux at the moment. I have no experience with Linux.  [Edit 2 times, last edit by Speedy51 at Jun 10, 2020 8:58:32 AM] |
||
|
|
hnapel
Advanced Cruncher Netherlands Joined: Nov 17, 2004 Post Count: 82 Status: Offline Project Badges: 
|
If it happens I'd like to vote for nVidia OpenCL 1.2 They will get there faster with CUDA, which is what Quarantine@Home is using for their Autodock project. https://quarantine.infino.me/ It looks like Nvidia has done the hard work for them: https://github.com/NVIDIA/nvidia-docker 'docker' as in a framework to run tasks in a semi virtualized environment has nothing to do with 'autodock' as in dock molecules to proteins... Also for the tasks to run on contributor PCs I don't think they will use the docker layer but just access the GPU directly, NVIDIA is helping out with the docker images to deploy programs that make use of cloud based GPUs. |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
Docker has to do with autodock if someone adapts it to do so. I will be happy if the current project develops their own app that accesses the GPU directly, but we will see when it gets here.
|
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges: 
|
They have been updating the OpenCL and CUDA code at the repository.
----------------------------------------Looks like they've been working quite a bit more on CUDA lately. If you check the commits, the first CUDA build is from March 31st but only recently did lots of the commits actually mention CUDA. https://github.com/ccsb-scripps/AutoDock-GPU Jim, I think Quarantine@home runs on OpenCL based on this tweet from the scientist who built the project. https://twitter.com/subC0smos/status/1262853097937711104  - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz [Edit 1 times, last edit by Falconet at Jun 16, 2020 9:09:35 PM] |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
Jim, I think Quarantine@home runs on OpenCL based on this tweet from the scientist who built the project. https://twitter.com/subC0smos/status/1262853097937711104 That is possible. I was basing my assumption on the fact that it runs only on Nvidia, and this line in the setup: sudo docker build -t quarantinegpu -f Dockerfile.gpu . && sudo nvidia-docker run -it -e ME=anonymous quarantinegpuAlso, to run OpenCl on Nvidia cards in Linux, you need to install the OpenCl drivers. They are not included in the standard drivers. But it is fast in any case. If they can get it to run on AMD, so much the better. But if they have to delay it six months, better go with what they have now and add the other stuff later. [Edit 2 times, last edit by Jim1348 at Jun 17, 2020 6:19:56 PM] |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges:
|
If they happened to have the CUDA version ready before the OpenCL version, they should launch as soon as possible. But I feel like they'll launch both versions at the same time.
----------------------------------------There are quite a few computers here at home but I would probably only run the GPU on my AMD Ryzen PC, a simple RX 550. The other more powerful GPU's would be the RX 580 on the Ryzen 5 1600 AF or the Nvidia GTX 1050 Mobile but those are not my computers and they don't run BOINC. Certainly not interested in paying the power cost of the RX 580 lol Probably not worth running the Vega 8 from the Ryzen 5 2500U laptop or any of the older laptops with GPU's. I'll see what I'll do once it launches. I hope we see BETA soon. - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
|
BladeD
Ace Cruncher USA Joined: Nov 17, 2004 Post Count: 28976 Status: Offline Project Badges: 
|
Is the OpenPandemics - COVID-19 Project working on a GPU port?
---------------------------------------- |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges:
|
Is the OpenPandemics - COVID-19 Project working on a GPU port? Yes. https://twitter.com/ForliLab/status/1261194223811887109 https://github.com/ccsb-scripps/AutoDock-GPU Edit: More confirmation https://www.worldcommunitygrid.org/forums/wcg...d,40665_offset,105#630059 - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz [Edit 1 times, last edit by Falconet at Jun 17, 2020 9:44:19 PM] |
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 865 Status: Offline Project Badges:
|
Falconet said:
----------------------------------------They have been updating the OpenCL and CUDA code at the repository. Looks like they've been working quite a bit more on CUDA lately. If you check the commits, the first CUDA build is from March 31st but only recently did lots of the commits actually mention CUDA. https://github.com/ccsb-scripps/AutoDock-GPU Jim, I think Quarantine@home runs on OpenCL based on this tweet from the scientist who built the project. https://twitter.com/subC0smos/status/1262853097937711104 Looking through Pull Request #82: jefflarkin said: This PR includes a post of AutoDock to CUDA, which enables running on Summit and has demonstrated a performance benefit over the OpenCL version on machine where both CUDA and OpenCL are available. A variety of changes are included: According to Wikipedia article Summit is an IBM-developed supercomputer ranked #1 on the Top500 publicly-known supercomputers in the world and is operated by the US Department of Energy at Oak Ridge National Lab. Runs Red Hat Enterprise Linux with a peak speed of 200 petaFLOPS. 9,216 POWER9 22-core CPUs 27,648 Nvidia Tesla V100 GPUs[1] The IBM POWER9 CPUs unfortunately won't be able to run AutoDock (that I know of...at least BOINC-related projects), but there may be use for those NVIDIA Tesla V100s. If so, having Summit working on OPN1 would dramatically speed up progress from years to hopefully weeks. We can only hope! Imagine if they created a WCG account so we can all stare at the stats. 
[Edit 1 times, last edit by hchc at Jun 18, 2020 11:40:44 AM] |
||
|
|
Aurum
Master Cruncher The Great Basin Joined: Dec 24, 2017 Post Count: 2391 Status: Offline Project Badges: 
|
I sure hope when they start GPU WUs they give them another, e.g. OPG1.
----------------------------------------This list makes one wonder why it's taking so long :-) https://insidehpc.com/2020/06/prace-awards-hu...hours-to-combat-covid-19/ Is there a way to benchmark a DC cluster?   ...KRI please cancel all shadow-banning |
||
|
|
|