|
Folding@Home is beta testing a new GPU folding core which supports both CUDA and OpenCL. It gives a speed boost on all Nvidia GPUs, especially on the lower core count GPUs: 1660 Super:31%, 1080Ti:25%, 2080Ti:20% or 3080:14%. Due to the quick return bonus, even though the relative speedup is higher on the lower tier cards, the actual points boost will be higher on the bigger cards. Looking forward to this going live so my 1660 Super folds faster than the RX 5600 XT. Unfortunately, low atom count work units still under-utilize the bigger cards, but they're working on segmenting GPUs better in the future, so they'll get work units more suited to their core count. Edit: I tried running BOINC on a couple of Android devices, but while I did return some valid results in the WCG Covid-19 project, I got a lot of work units that failed due to segfaults, so I gave up. It was a cheap phone and a cheap Asus pad, so they might have had instability in anything from Android to the storage. Vir fucked around with this message at 12:44 on Sep 28, 2020 |
# ¿ Sep 28, 2020 11:02 |
|
|
# ¿ Apr 19, 2024 23:31 |
|
Folding@Home has released the CUDA folding core. There are also some bug fixes for AMD cards, but with this the Nvidia cards have an even bigger advantage for folding.
|
# ¿ Sep 29, 2020 10:56 |
|
Zarin posted:Yeah it's super dumb (in a good way!). I went from 540k PPD to over 1.05m overnight. Absolutely silly. AMD's engineers better get working on those ROCm drivers and see if they can optimize them for folding without hurting the more memory intensive compute tasks. Zarin posted:Is there any reason why I *shouldn't* run Folding@Home on my CPU? CPU folding is less energy efficient than GPU folding, so if you need to run your air conditioner, then you should probably pause CPU folding until the weather gets colder. Even so, Folding@Home needs CPU folders, so CPU folding isn't dead. If you have a fairly modern Ryzen or Threadripper CPU with SMT, the most efficient number of threads appears to be 2 below the CPU's thread count, and subtract 2 threads per GPU. So if you're only CPU folding on a 32 thread Ryzen 9 3950X, the most efficient setup is folding on 30 threads. If you have one GPU as well, then 28 or 26 CPU threads is better. A Ryzen 5 3600 running a GPU is best run with 8 threads. CUDA might have changed this, so we might to check. The CUDA core still shows 100% CPU utilization, but this is just an active wait state, so real testing is needed to see if you can now fold with more CPU threads together with the GPU.
|
# ¿ Sep 29, 2020 20:49 |
|
If you check the logs, you'll probably see many instances of the folding core automatically dropping down to 8 folding threads because it doesn't like a number of threads which is divisible by 5, 7 or other large primes. Edit: Your CPU has AVX2, which means it should be pretty good for folding.
|
# ¿ Sep 29, 2020 21:08 |
|
That's weird, which OS? If you're already maxed out doing BOINC, it might not be worth the time to set up Folding as well, but I've never heard of hibernation being interrupted by folding. I do know that the "on idle" setting has very mixed results, especially with GPU folding, and that sometimes manually pausing the GPU folding or making a script for it works better than idle detection. Folding is actually a bit time sensitive, especially the current "Moonshot" project sprints, which only gives you like 24 hours to complete a work unit. So if you don't use your computer most every day, folding might not be suited for it.
|
# ¿ Sep 30, 2020 17:15 |
|
You might try to set the "disable-sleep-when-active" to "false" in the Expert tab in your client configuration. That would allow you to use Sleep, but you should check if the GPU core correctly resumes from checkpoint when it wakes up again, or if it just abandons the work unit. The reason for this issue seems to be that Sleep and Hibernate doesn't store the contents of the GPU's VRAM, only the state of CPU processes, so FAH "solves" it by blocking the sleep function while GPU folding. It's apparently hard to program around too, because of the GPU permissions that Windows gives to applications. Same reason why you shouldn't be folding on Remote Desktop, but should be physically logged in. (You can start new GPU work units while logged into Remote Desktop, but if you log out it won't start a new one when the current WU finishes.)
|
# ¿ Sep 30, 2020 19:54 |
|
BurritoJustice posted:now when I go to fold the CPU is running around the same but the GPUs are sitting on like, 10k each. Which is obviously not worth it for the electricity used. Does anyone have any ideas? If you return a work unit after Timeout, but before Expiry, you will still get some credit but the WU will be re-issued to someone else. And if you return before Timeout, your QRB is higher the faster you return it. COVID-19 Moonshot work units have especially short timeouts because they are so time-critical. Researchers are using the data to prioritize chemical compounds to synthesize and assay in-vitro. More traditional projects have longer timeouts on them. Can you try pausing the CPU and see if the GPUs start working again? Two CPU threads should be dedicated to feeding the GPUs with work and sanity-chekcing their results. If your PPD permanently dropped, then I'd suspect some kind of hardware issue. Have you asked in the foldingforum.org forums?
|
# ¿ Oct 10, 2020 14:28 |
|
Folding work units are not supposed to be done by two folders. The only cases where a work unit is issued to more than one folder is if it passes timeout, it crashes, gets dumped or there's something wrong with the work server. The servers validate the work on their own, because it takes relatively little computation to validate the folding work once it has been done.
|
# ¿ Oct 11, 2020 07:27 |
|
Folding@home has quietly released a new client version (7.6.20), and there is now an ARM Linux / Raspberry Pi version as well: https://foldingathome.org/alternative-downloads/ e: Also they posted some science news: https://foldingathome.org/2020/10/20/covid-moonshot-sprint-4/ quote:Just as we were doing this, something extremely unexpected happened: The chirally-separated version of the compound we started with turned out be 100 nM! That’s 25 times more potent (meaning you need 25x less drug to get the same effect in shutting down the protease) than what we thought we started from! Vir fucked around with this message at 11:39 on Oct 20, 2020 |
# ¿ Oct 20, 2020 11:22 |
|
Folding at Home is running a 20 year anniversary stream on Twitch: https://twitch.tv/videos/807519293 e: Link to replay. Vir fucked around with this message at 13:34 on Nov 19, 2020 |
# ¿ Nov 18, 2020 19:20 |
|
There is a security vulnerability that affects those who use the Folding@Home GUI prior to version 7.6.20 to control remote folding instances over insecure networks: https://foldingathome.org/2020/11/23/update-for-those-using-advanced-remote-client-management-configurations/ Basically, the folding client or an attacker on the folding client's network could trick the GUI into running arbitrary code. This doesn't matter to those who fold on their own computers, but maybe if you're running folding on your webserver or something and just opened it up to the world. mdxi posted:Sometime in the past few days I hit 100 years of CPU time for WCG's Mapping Cancer Markers project
|
# ¿ Nov 24, 2020 11:10 |
|
I fold on my game computer while I'm not gaming, and fold on some e-waste machines for additional heating in the winter time. Tuxide posted:In Folding@home, we are still #68 in rankings. Its compute power is now about 240 petaFLOPS, a shadow of the 2.62 exaFLOPS it was back towards the end of April.
|
# ¿ Dec 6, 2020 01:00 |
|
Dr. Bowman shares some of the F@H results so far. They are moving into animal testing of candidates for therapeutic compounds. https://www.youtube.com/watch?v=6w8xb__A8Gc
|
# ¿ Dec 8, 2020 00:01 |
|
Folding@Home simulations might have found a way to disable a bunch of coronaviruses, like SARS-CoV-2 (Covid-19), SARS-CoV-1 and MERS. They might even have found a therapeutic for the common cold. https://foldingathome.org/2020/12/16/sars-cov-2-nsp16-activation-mechanism-and-a-cryptic-pocket-with-pan-coronavirus-antiviral-potential/
|
# ¿ Dec 19, 2020 17:18 |
|
mdxi posted:That is work right there. Protein engineering is getting really, really interesting. Here's a status update on this. They hope to have a clinical candidate this month. https://www.youtube.com/watch?v=fH0hIXC3bUg
|
# ¿ Mar 12, 2021 15:04 |
|
Will this allow MacOS and ARM Linux users to use GPUs for Open Pandemics?
|
# ¿ Apr 1, 2021 19:20 |
|
There are some old Mac Pros with discrete AMD GPUs in them, and even some from Nvidia before that, but Folding@Home doesn't support any of them because there was a bug in OpenCL for MacOS back when it was more worthwhile, and MacOS has deprecated OpenCL and replaced it with Apple's own Metal API.
|
# ¿ Apr 2, 2021 00:04 |
|
The F@H statistics page has been given a makeover. It now displays team logos for those that have it, and it has also broken the client's web frontend. SAGoons is ranked 69, but lacks a logo: https://stats.foldingathome.org/team/150 Old page: https://statsclassic.foldingathome.org/team/150
|
# ¿ Apr 12, 2021 00:43 |
|
F@H has rolled back the new stats pages, because it broke the web client. This premature rollout also revealed that many third party websites were hammering the statistics API with excessive calls. F@H has only one programmer on staff, which I guess is a typical symptom of how research grants pay for PhD projects, but infrastructure and support functions are under-funded. Here's the beta page: https://statsbeta.foldingathome.org/team/150
|
# ¿ Apr 13, 2021 09:21 |
|
John Chodera from Folding@Home posted a status update about the Covid Moonshot:quote:To very quickly update where we are:
|
# ¿ Jun 14, 2021 15:01 |
|
mdxi posted:Here, with proper central HVAC, the irony is that the nodes are now off in a bedroom that I'm using as my office and while airflow is fine, this room becomes a hotspot that the centrally-located thermostat doesn't care about.
|
# ¿ Aug 25, 2021 12:42 |
|
The Covid Moonshot (of which Folding@Home is a part) is getting a 8 million GBP grant from Wellcome and the Covid-19 Therapeutics Accelerator to develop a treatment for coronaviruses. Press release: COVID Moonshot funded by COVID-19 Therapeutics Accelerator to rapidly develop a safe, globally accessible and affordable antiviral pill News post on Folding@Home: https://foldingathome.org/2021/09/27/covid-moonshot-wellcome-trust-funding/?lng=en John Chodera posted:This funding will enable the Moonshot to rapidly complete its final stages of lead optimization and perform the preclinical studies needed to reach the equivalent of Investigational New Drug (IND) filings to begin clinical trials.
|
# ¿ Sep 28, 2021 09:31 |
|
I know what you mean about AMD GPU support - I messed around a lot with that in Linux until my AMD card became obsolete in terms of PPD. Folding@Home has beta support for Intel GPUs. So far, some projects can be folded on integrated Intel GPUs, but the reason they're even bothering with this support is in anticipation of higher powered discrete Intel GPUs.
|
# ¿ Sep 29, 2021 19:09 |
|
This post contains speed gains for various cards with CUDA in regular and Moonshot Work Units: https://foldingathome.org/2020/09/28/foldingathome-gets-cuda-support/?lng=en
|
# ¿ Sep 29, 2021 22:28 |
|
AMD cards are really bad at Folding@Home right now, due to recent issues with bad drivers on top of non-optimized OpenCL. AMD cards are passable for gaming use, but under-perform in protein folding. It's not like AMD needs to mend this to sell more cards - cryptocurrency miners buy them up anyway.
|
# ¿ Nov 2, 2021 15:36 |
|
Folding@Home is running another Covid Moonshot sprint to assess the potency of candidate compounds for antiviral medications. You can expect GPUs to see higher point yields while the sprint is running. More information on this, and where it fits into work on other antivirals, can be read here: https://foldingathome.org/2021/12/27/covid-moonshot-sprint-11/?lng=en
|
# ¿ Dec 28, 2021 23:32 |
|
On Folding@Home, some runs of the Covid Moonshot sprints had some errors in it that would manifest as several "Force RMSE error" and dumped WUs. The researcher has withdrawn those now, and prepareing a new Covid Moonshot sprint. You might want to check that your GPU hasn't been stuck in a "FAILED" state - if they have, restarting the Folding@Home client should take care of it.
|
# ¿ Jan 15, 2022 10:45 |
|
An update from an Alzheimer related folding project at Folding@Homejjmiller posted:One more note on this project. We've finished a first pass analysis on the data we collected for p18202 and have written a paper detailing this analysis and findings. While this paper is still in peer review for official publication, we've made the paper available via the biology preprinting server, Biorxiv. You may find the paper here: https://www.biorxiv.org/content/10.1101/2022.02.02.478828v1.Here are the highlights-
|
# ¿ Feb 4, 2022 16:36 |
|
Since AMD is turning away from OpenCL support, and their recent drivers have been performing worse in OpenCL compute than older drivers, it's good news that OpenMM, which is used as the GPU compute core in Folding@Home seems to be getting support for HIP, AMD's own compute API. I don't yet know if this will also be integrated into Folding@Home, but I see no reason why it shouldn't. It's still a bit sad that AMD, unlike Nvidia, haven't provided any development support for making Folding@Home run efficiently on their GPUs, but I guess they have too much work to do elsewhere for paying customers.
|
# ¿ Feb 8, 2022 22:29 |
|
Yes, it used to be that AMD had the best driver support on Linux, right? I'm not sure they're so bad at graphics, but if you're using AMD for compute, even in Windows, you'd better be using an old driver. I would have thought that AMD's interest in HPC would cause them to be more on the ball regarding compute driver support, but here we are.
|
# ¿ Feb 9, 2022 10:53 |
|
mdxi posted:My understanding is that Nvidia's OpenCL driver also sucks, though perhaps not quite as much as AMD's. While AMD haven't dropped OpenCL completely from their drivers, support for OpenCL has been dropped from Radeon Rays and Blender Cycles X, so they might no longer be validating OpenCL performance. It's likely that they'll drop OpenCL support eventually.
|
# ¿ Feb 10, 2022 10:11 |
|
Vir posted:Nvidia's drivers haven't regressed in OpenCL performance like the AMD drivers have. Edit: A new Covid Moonshot has started: quote:Projects 13462 and 13463 are moving to full FAH for COVID Moonshot Sprint 12 shortly, which aims to help prioritize the final stages of synthesis for the first-generation SARS-CoV-2 focused compounds for the COVID Moonshot. Vir fucked around with this message at 01:59 on Feb 15, 2022 |
# ¿ Feb 15, 2022 00:59 |
|
I've never heard of one. The problem is going to be bandwidth. If she wants to do a community service she could sign up as a Bittorrent mirror for some obscure scientific content, open source distributions, etc. - in other words legal content. She could even earn "torrent-coins" for hosting, with some clients. For example, some of the content on The Internet Archive has torrent downloads, and you might want to seed those files to help other people download them faster. Another thing she might do is digitize movies and other cultural artefacts which aren't in the public domain for preservation for the future, but she can't legally open that up for other people.
|
# ¿ Jul 17, 2023 14:15 |
|
|
# ¿ Apr 19, 2024 23:31 |
|
Congrats on 400 CPU-years. I'm still running FAH whenever the heat is useful, or on days when we've had negative electricity prices. I'm still using my 1660 Super and a 1080 - never felt the need to upgrade yet.
|
# ¿ Aug 22, 2023 17:17 |