| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: Work unit availability |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 822
|
|
| Author |
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
Actually i would really like World Community Grid to increase GPU work units for Open Pandemics and start GPU work units for the other projects and any new projects that will come out in the future. WCG has absolutely no control over what platform (CPU/GPU) is used to do research here. It is 100% controlled by the researchers. Contact the researchers for each project and submit your suggestions to them keeping in mind that most of the type of research that is done here is not conducive to running on GPUs. That’s why only 2 of 31 projects here have been ported to GPUs.
In 1969 I took an oath to defend and protect the U S Constitution against all enemies, both foreign and Domestic. There was no expiration date.
  |
||
|
|
yoerik
Senior Cruncher Canada Joined: Mar 24, 2020 Post Count: 413 Status: Offline Project Badges: 
|
A quick note about workunit availability for the project. As noted in the post about the outcome of the stress test the researchers have a limit on how fast they can process results and make use of them. We are currently working with them to target about 500 batches per day (this is combined between CPU and GPU). It is a little tricky to manage to this because even though there is a consistent number of ligands per batch, the number of workunits created per batch varies based on the estimated difficulty of the ligands. Thus we have to tweak the rate we send out workunits. Right now the 7 day average of batches for GPU is 727 batches/day and CPU is 185 batches/day. This is above the 500 batches/day goal. As a result, we reduced the GPU pace 2 days ago from 3000 workunits/day to 1700 workunits/day. We have been attempting to reduce the pace of the CPU work as well. Unfortunately, there is a fairly large set of users who only have OpenPandemics selected and thus we are not able to shift as much of the CPU power to our other projects as I had hoped. We will continue to adjust things where we can to support the 500/batches per day goal and to shift work to GPU as much as possible without making the CPU work intermittent. I would ask that if you are someone who has only OpenPandemics selected, please consider adding Mapping Cancer Markers and/or the African Rainfall Project in addition to OpenPandemics. OpenPandemics will continue at the same pace regardless of how much CPU power is allocated to it as the GPU side will pick up any slack left over from the CPU side. Thank you for the additional information. Hopefully more users follow your request - but perhaps a news post would be more worthwhile - given it'll reach more users than buried in the forums.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I will remove 240 threads from OPN as I have no interest in the project. Will focus on MIP and ARP until MIP ends. Stopped OPNG when no tasks were available.
----------------------------------------However, don't quite understand why a group would put forth an effort to create a GPU version if you don't have the staff to take advantage of it. All that trouble only to throttle it back to CPU speeds. What a waste.... [Edit 1 times, last edit by Former Member at Jun 12, 2021 3:23:56 AM] |
||
|
|
Dayle Diamond
Senior Cruncher Joined: Jan 31, 2013 Post Count: 452 Status: Offline Project Badges: 
|
Individually removing our computers from Open Pandemics CPU tasks is pointless. The tasks are created and will be assigned to somebody. If you don't waste your threads, somebody else will.
----------------------------------------Knreed wrote about avoiding OPN becoming an intermittent project. I thought that was a curious thing to say. Maybe feast-or-famine tasks are harder on the system, or harder to manage, or even result in too many forum posts to moderate as people discuss getting work, constantly bicker, and troubleshoot out-of-work messages? In every recent project update, there is less than a month of MCM batches in the queue. I've been saying from the start that we need to shift 100% of this project to the GPUs to free up massive CPU computation for other projects. Now that the Microbiome Immunity project is ending, we're talking 280 years of crunching power a day, with one project big enough to absorb the demand. But maybe it's not big enough; maybe that would be enough to make MCM batches become intermittent? Is that a problem worth avoiding? Would WCG staff, or crunchers mind if every project backlog ran dry? Do we mind so much as to be willing to run millions of low priority work units to keep the supply stable? [Edit 1 times, last edit by Dayle Diamond at Jun 12, 2021 5:35:44 AM] |
||
|
|
yoerik
Senior Cruncher Canada Joined: Mar 24, 2020 Post Count: 413 Status: Offline Project Badges:
|
the point would be to encourage computers only crunching OPN1 to begin accepting WUs from other projects as well, Dayle? So the grid can shift more OPN WUs over to GPU, and not completely lose or waste the CPU capacity, by carrying it over to other projects?
----------------------------------------That's why I pointed out that it might be a good idea to send out a news post to everyone, encouraging people to select multiple projects. It's a delicate situation - there is no perfect solution. However I think it's a stretch that *every* project backlog runs dry. As of today, only a few dozen crunchers at most will read this message in the forums - and even if it is just in sum a few hundred threads that we transfer over to another project, that can make some sort of impact by reducing CPU's demand for OPN WUs - by letting those WUs flow to people who *only* selected OPN1? It's marginal, but it's something - and the best we as individual crunchers can do right now? |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges: 
|
I basically haven't run OPN since I got my 5 year badge in July 2020 (or so). At the time, I was waiting for the GPU app. I ran a few OPNG tasks but mostly on Google Colab.
----------------------------------------I'm 100% in favour of shifting as much OPN to the GPU app as possible. Given OPN has a GPU app available, efficiency is far greater and the CPU is far more valuable on other projects, as others have mentioned. I had noticed OPN was running far less tasks and CPU years per day while other projects were getting a higher share. Now I know why. A notice/news article is a good idea. Keep it up.  - 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 2 times, last edit by Falconet at Jun 13, 2021 11:06:10 AM] |
||
|
|
GunZi
Senior Cruncher Joined: Apr 1, 2020 Post Count: 160 Status: Offline Project Badges:
|
I'm only 60 days away from the 10 year badge, then I'll switch to the mapping cancer markers project.
----------------------------------------[Edit 1 times, last edit by GunZi at Jun 12, 2021 11:23:02 AM] |
||
|
|
mgpointner
Advanced Cruncher Argentina Joined: Nov 16, 2009 Post Count: 55 Status: Offline Project Badges:
|
I'm only 60 days away from the 10 year badge, then I'll switch to the mapping cancer markers project. I'll switch to the mapping cancer markers project too  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Right now i would say that i am getting a lot less CPU work units for Open Pandemics as over 90% of my new work units are for MCM and that is happening on both computers
|
||
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges: 
|
I've thought right from the start of the AD-GPU work for OPN that it would be best to discourage use of CPUs for the project. However, I remember reading at the time that WCG considered the CPU work very important too, as it would be used for docking the smaller ligands. But there was no mention of why the smaller ligands could not be run on GPUs instead. Knreed's post above indicates that the policy of wishing to dock the smaller ligands only on CPUs has now changed.
So, I have duly tweaked my WCG Device Profiles to disable downloading OPN CPU work or limit the number of WUs to 4 or fewer. At the same time, I set MCM and MIP to receive unlimited WUs. My work caches are set to 0.6 days' work. Now most of my devices, including those that have never run AD-GPU, have downloaded crazy amounts of MCM and MIP WUs, like 743 x MCM plus 237 x MIP in one case. Previously, allowing unlimited OPN1 WUs in my Device Profiles did not result in this behaviour. What is going on? |
||
|
|
|