| 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: 89 Status: Offline Project Badges: 
|
Just for fun i ran two concurrent WU on an Intel HD4000. Took 2+ hours to complete (while running 4/4 cpu jobs and 2 NVIDIA) , none validated yet so no idea of the points that might be awarded...
---------------------------------------- |
||
|
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1259 Status: Offline Project Badges: 
|
2. The official name for this is going to be opng, the work units for GPU will only be taged with OPNG in their name. I think others have asked this question and I haven't come across an answer as yet. When OPNG application goes live will the statistics returned be separated or will they be combined with CPU work? Pleased to hear there were a good number of virtual high-fives with the researchers. Uplinger is your time zone UTC/GMT -5 or 6? On Sunday morning I will be UTC +12 currently I am UTC +13  |
||
|
|
pokemonlover1234
Cruncher Joined: Mar 4, 2021 Post Count: 26 Status: Offline Project Badges: 
|
Glad to see the beta is going generally well! Looking forward to dedicating my GPU to OPN instead of trial factoring primes in its spare time. When is this expected to go into production, if you dont mind me asking?
---------------------------------------- |
||
|
|
ca05065
Senior Cruncher Joined: Dec 4, 2007 Post Count: 325 Status: Offline Project Badges:
|
I tried running this beta on a Radeon HD6700 series GPU. The work unit did run - initially at 10% per minute at 10% progress before slowing to 1% per minute at 20% progress after 20 minutes and 0.5% per minute at 99.999% progress after 2 hours with zero time remaining.
----------------------------------------After a suspend / resume the work unit restarted at 0% progress. I aborted the work unit. Should such GPUs be excluded from being sent these work units? Added further information: In my stderr file in results status the job #0 had not completed. The wingmen completed 39 jobs in 0.03 and 0.05 hours. [Edit 1 times, last edit by ca05065 at Mar 31, 2021 11:15:28 PM] |
||
|
|
hnapel
Advanced Cruncher Netherlands Joined: Nov 17, 2004 Post Count: 82 Status: Offline Project Badges: 
|
I had a blast with the new beta batch! On some systems I also have Folding at home running on the GPU, so I needed to use the device control to decide which computers could or could not run the BOINC GPU tasks. I happen to have 2 Nvidia RTX 2080 Super GPU's, the beta jobs take about 2 minutes on each of those(!), that's about 60 times faster than a CPU job on the same computer. Given the fact I can reasonably run 12 CPU jobs at the same time, a rough calculation shows my GPU's will be able to process 10 times as much work as the main CPU (not even mentioning the jobs for the Intel HD graphics, which take about 30 minutes to complete). To keep the GPUs going I had to configure the setting 'keep extra work' otherwise the jobs were completed before new ones arrived.
I noticed the GPU utilization on the Supers was going up and down all the time (checking with GPU-Z), alternating between 99% utilization and very low values like 1-4% (the graph showed a saw figure), so this means the efficiency is subpar, but I do not mind since it keeps the GPUs temperature down, if it was at 99-100% all the time I would probably not be able to run them all the time. |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
I am sorry to having to jump in here but all my Beta-WUs are ending with Valid but Claimed/granted 0.00/0.00. Is there a place I can look to see why this is? ........ It looks good to me, but no points are displayed. Not that I care too much about Points, but I'm just wondering. Cheers aend Aend, This looks like a work unit that was sent before the updated validator was in place. I have mentioned above, that I regret that I can not retroactively provide points to these previous work units. This is why the granted points show zero. Thanks, -Uplinger |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
I had a blast with the new beta batch! On some systems I also have Folding at home running on the GPU, so I needed to use the device control to decide which computers could or could not run the BOINC GPU tasks. I happen to have 2 Nvidia RTX 2080 Super GPU's, the beta jobs take about 2 minutes on each of those(!), that's about 60 times faster than a CPU job on the same computer. Given the fact I can reasonably run 12 CPU jobs at the same time, a rough calculation shows my GPU's will be able to process 10 times as much work as the main CPU (not even mentioning the jobs for the Intel HD graphics, which take about 30 minutes to complete). To keep the GPUs going I had to configure the setting 'keep extra work' otherwise the jobs were completed before new ones arrived. I noticed the GPU utilization on the Supers was going up and down all the time (checking with GPU-Z), alternating between 99% utilization and very low values like 1-4% (the graph showed a saw figure), so this means the efficiency is subpar, but I do not mind since it keeps the GPUs temperature down, if it was at 99-100% all the time I would probably not be able to run them all the time. In theory, we will be getting much more difficult jobs within a work unit. Right now they are still very much similar to CPU with minor changes. Once we get more into the GPU work, we will see longer times used for each job. The reason it goes down to zero during that time is because it is unloading and loading the next job to run. During this switch between jobs, it writes the outputs from the jobs. Basically it is exiting to CPU mode and then entering GPU mode each job. Thanks, -Uplinger |
||
|
|
widdershins
Veteran Cruncher Scotland Joined: Apr 30, 2007 Post Count: 674 Status: Offline Project Badges: 
|
You could run MSI aftrburner. It works on non MSI cards as well (just without all the extra bells and whistles for overclocking). It allows you to set a target GPU temp as the priority, or power usage, and set a custom fan speed profile.
So you can tweak things to keep the temp lower but have the gpu still fully utilised, by allowing the software to throttle back the clock speed when it gets over the target temp, or by running the fan up to higher RPM. |
||
|
|
widdershins
Veteran Cruncher Scotland Joined: Apr 30, 2007 Post Count: 674 Status: Offline Project Badges:
|
Ohh uplinger is back on, must check for new WU's arriving.
 |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
I noticed the GPU utilization on the Supers was going up and down all the time (checking with GPU-Z), alternating between 99% utilization and very low values like 1-4% (the graph showed a saw figure), so this means the efficiency is subpar, but I do not mind since it keeps the GPUs temperature down, if it was at 99-100% all the time I would probably not be able to run them all the time. The variance in GPU utilization comes during the time one job of the WU has completed and the next one starts. Perfectly normal if you're only running 1 task at a time. Run more than 1 task at a time and you probably won't see that.
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.
----------------------------------------  [Edit 2 times, last edit by nanoprobe at Apr 1, 2021 12:40:55 AM] |
||
|
|
|