| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: OpenPandemics GPU Beta Test - March 26 2021 [ Issues Thread ] |
| No member browsing this thread |
|
Thread Status: Locked Total posts in this thread: 511
|
|
| Author |
|
|
Jorlin
Advanced Cruncher Deutschland Joined: Jan 22, 2020 Post Count: 90 Status: Offline Project Badges: 
|
Got my problem with stuck Jobs sorted out (short: Acronis Update, "Active Protection" got reenabled and somehow forgot to notify me it was blocking stuff).
----------------------------------------Everything running smoothly now on GTX1050Ti, GTX960, Intel HD4000 :)  |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
@Jorlin, thanks for the update and what was causing it. This was something I was worried about as I could not recreate this problem. So it sounds like 3rd party security software was stalling the jobs?
@Ulrich Metzner, I do not know why, but it seems like the GT cards are generally on my naughty list of graphics cards that can't work. I will need to do some research to see what the differences are between GT and GTX cards. @cpalmer, I'm guessing where you are getting the performance increase in during the time between ligands. What this means is that when it is switching between ligands the science application unloads what was run on the GPU and prepare the next ligand to be pushed to the GPU. With these still being rather short, setting to 2 seems like it might fill in the gaps, I'm not sure what the consequences are going to be when the work units get more difficult. Thanks, -Uplinger |
||
|
|
Mamajuanauk
Master Cruncher United Kingdom Joined: Dec 15, 2012 Post Count: 1900 Status: Offline Project Badges: 
|
Hi Uplinger
----------------------------------------I've been seeing tasks running for a short time 0.03 seconds with a Valid result and NO credit!
Is this correct/to be expected as the time is so short? Thanks
Mamajuanauk is the Name! Crunching is the Game!
  |
||
|
|
Jorlin
Advanced Cruncher Deutschland Joined: Jan 22, 2020 Post Count: 90 Status: Offline Project Badges:
|
@Jorlin, thanks for the update and what was causing it. This was something I was worried about as I could not recreate this problem. So it sounds like 3rd party security software was stalling the jobs? Thanks, -Uplinger Acronis is a backup software but they built in malware protection at some point so YES. I turned it off and forgot about it. The update reenabled it. I now excluded the BOINC folder before turning it off again... just in case |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
Hi Uplinger I've been seeing tasks running for a short time 0.03 seconds with a Valid result and NO credit!
Is this correct/to be expected as the time is so short? Thanks So the points are still an issue, I am working towards fixing that, but I am planning to credit time based on elapsed time and not CPU time as that makes more sense for the GPU project. That would be the 0.37 hours you see on your report. Note: I am not going to hold up the release of GPU over points...it is on my priority list near the top and I would like to have it fixed before launching. Thanks, -Uplinger |
||
|
|
PMH_UK
Veteran Cruncher UK Joined: Apr 26, 2007 Post Count: 774 Status: Offline Project Badges:
|
Are there limits on workunits per PC?
----------------------------------------I have 50 on 1 PC. Paul.
Paul.
|
||
|
|
zdnko
Senior Cruncher Joined: Dec 1, 2005 Post Count: 229 Status: Offline Project Badges: 
|
4 nv on GTX745 (2 wu same time) in 35min (every wu)
----------------------------------------1 int on HD4600 in 40min 5 of 5 valid! i7-4790 Windows 10: Core x64 Edition, (10.00.19041.00) [Edit 1 times, last edit by zdnko at Mar 26, 2021 10:36:04 PM] |
||
|
|
Jorlin
Advanced Cruncher Deutschland Joined: Jan 22, 2020 Post Count: 90 Status: Offline Project Badges:
|
I'm not here for the points and the shiny little badges are based on elapsed time ;)
----------------------------------------Seriously, if i was after points to show off on BOINCstats, i wouldn't be crunching here anyway. Would i like to get points? sure. But i don't know how to calculate them fairly, never felt the need to dig into it. Based on elapsed time? I'll run four WU on a single GPU and quadruple my points. The other GPU projects i ran (Einstein and Primegrid) give fixed points for any completed task. |
||
|
|
goben_2003
Advanced Cruncher Joined: Jun 16, 2006 Post Count: 146 Status: Offline Project Badges:
|
I got a few WU's which all worked fine on my GTX 1660 and finished under 3 minutes. It has been already reported earlier but one thing that worries me a bit is that the GPU work level quickly shifts from 0 to 100% all the time. I hope that that doesn't have an adverse effect on the life expectancy of my GPU. Hi Drago75: I think we talked about this in the previous beta threads. Unfortunately at this time it is worked around at the per machine level. My RTX 2060m is only 26% avg load with 1 work unit, so I set it to run multiple. You can put something like this in your app_config.xml to run multiple. Set the gpu_usage to 1/N where N is the number you want to run at the same time. Here is with 2 running at the same time. Try that and then increase N by 1 until the usage looks good. While N are actually running -- during the beta you may have trouble getting enough units for N to actually run at the same time. Allocation of the CPU to handle the GPU tasks can also be adjusted independently of GPU usage. The below means that if there are 2 units, boinc will count 1 of your available cpu threads to be taken by the GPU, so 1 less CPU work unit will run during that time. <app_config> Edit: corrected filename  [Edit 1 times, last edit by goben_2003 at Mar 26, 2021 10:38:04 PM] |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
Are there limits on workunits per PC? I have 50 on 1 PC. Paul. The beta mechanism should limit it to 1 per core of the machine for CPU. I have not confirmed this, but it may be sending you 1 per thread of CPU for GPU. If you have 2 GPU's, say an intel and nvidia, then it is probably sending you up to 1 per thread per gpu, so twice as much as normal BETA but not taking into account the GPU either. I know kind of confusing, but I did not see this as a bad thing for this beta for a host to get more than 1 per GPU, as it has given those a chance to over schedule their own GPU for this application to see what it would do. Thanks, -Uplinger |
||
|
|
|