| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: OpenPandemics - GPU Stress Test |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 781
|
|
| Author |
|
|
hnapel
Advanced Cruncher Netherlands Joined: Nov 17, 2004 Post Count: 82 Status: Offline Project Badges: 
|
I threw in my GT 710 (on Linux) for good measure.
|
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
and actually, bigger tasks is what they're pumping out with these high batch number tasks. anything 13,000+ it seems. they're running up to 10x longer than the 7000 and under batches on my 2080ti. (13,000+ = ~9 minutes, 7000- = ~1min or less) none have validated yet (pending), the question is, will they receive 10x credit as the older tasks? my gut feeling says no. uplinger has mentioned in a previous posts that the WUs have some maximum credit reward baked in, and that actual credit granted is based on the percentage of calculations needed to get the result. NOT the total calculations. and both old and new tasks are tagged with the same estimated flops size (incorrectly) at 31,450 GFlops, so my guess is that the maximum credit is the same as well. Max points per task is 1800. The two validated 5 digit tasks I had on 4/25 were given high 1600s low 1700s. I am reviewing what I have in the database. If a point adjustment is needed, I will make that shortly. Thanks for the feedback on them being monster jobs in length of time. I am reviewing why that is the case, because these are rigid target/ligand jobs and those generally run faster. But we are combining MANY more in a batch because the ligands are small. Might be a combination of a few factors. Thanks, -Uplinger |
||
|
|
Grumpy Swede
Master Cruncher Svíþjóð Joined: Apr 10, 2020 Post Count: 2161 Status: Offline Project Badges: 
|
The validator doesn't seem to be interested in validating these larger WU's from batches 13345 - 41773. The previous lower batches were validated very soon after they were finished. No wingman, I'm _0 on all of these WU's, but the validator isn't interested in even trying to validate. Edit: Why does this "large" WU start at "job" #56? Never seen that before either. All other previous WU's always started at job #1 https://www.worldcommunitygrid.org/ms/device/...og.do?resultId=1657133303 It shows starting at 56 because BOINC only uploads the last X bytes in the stderr to us. This limits what you see on the website. As for the validator doing well, I have bumped up to 8 cores and I'm monitoring where it is at. Thanks, -Uplinger Thanks for the quick answer Uplinger! Now I know the answer to the #56 question. When it comes to the validator though, it does validate the older WU types pretty quick, but just shows the middle finger  at the "larger" ones. We'll see if you can persuade it to start behaving |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges:
|
Please note:
We still are tweaking the load balancer for optimal performance. This seems to have been the root cause of all the troubles the past 12 hours. We believe we have a fix in place, but would like to try a few other options over the next little bit to make things even smoother. Currently the load on the download/upload server is pretty calm... Thanks, -Uplinger |
||
|
|
lakotamm
Cruncher Joined: Jul 4, 2019 Post Count: 5 Status: Offline Project Badges:
|
I am getting some very weird behavior here. The GPU seems to be loaded only very little. I have Ryzen 7 5800H (already running other tasks on both iGPU and CPU) and RTX3060. Is this normal? Thanks!
----------------------------------------https://i.ibb.co/7VDLfRR/Screenshot-2021-04-27-165859.png [Edit 1 times, last edit by lakotamm at Apr 27, 2021 3:04:29 PM] |
||
|
|
m0320174
Cruncher Joined: Feb 13, 2021 Post Count: 11 Status: Offline Project Badges:
|
I am getting some very weird behavior here. The GPU seems to be loaded only very little. I have Ryzen 7 5800H (already running other tasks on both iGPU and CPU) and RTX3060. Is this normal? Thanks! https://i.ibb.co/7VDLfRR/Screenshot-2021-04-27-165859.png I noticed that those bigger GPU workunits also consume considerable more CPU cycles (not only absolute but also relative) during which your GPU is basically waiting. Example: https://ibb.co/WDKKJWd ==> this workunit took 8 minutes to complete, during the first 3 minutes the GPU was not doing anything I also saw some cases in which the GPU was idling in the middle of the workunit. [Edit 2 times, last edit by m0320174 at Apr 27, 2021 3:29:39 PM] |
||
|
|
spRocket
Senior Cruncher Joined: Mar 25, 2020 Post Count: 274 Status: Offline Project Badges:
|
Uh oh... just tried to turn in some WUs and the connection failed...
----------------------------------------EDIT: False alarm... it retried on its own and it worked. [Edit 1 times, last edit by spRocket at Apr 27, 2021 3:26:34 PM] |
||
|
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: 
|
and actually, bigger tasks is what they're pumping out with these high batch number tasks. anything 13,000+ it seems. they're running up to 10x longer than the 7000 and under batches on my 2080ti. (13,000+ = ~9 minutes, 7000- = ~1min or less) none have validated yet (pending), the question is, will they receive 10x credit as the older tasks? my gut feeling says no. uplinger has mentioned in a previous posts that the WUs have some maximum credit reward baked in, and that actual credit granted is based on the percentage of calculations needed to get the result. NOT the total calculations. and both old and new tasks are tagged with the same estimated flops size (incorrectly) at 31,450 GFlops, so my guess is that the maximum credit is the same as well. Max points per task is 1800. The two validated 5 digit tasks I had on 4/25 were given high 1600s low 1700s. I am reviewing what I have in the database. If a point adjustment is needed, I will make that shortly. Thanks for the feedback on them being monster jobs in length of time. I am reviewing why that is the case, because these are rigid target/ligand jobs and those generally run faster. But we are combining MANY more in a batch because the ligands are small. Might be a combination of a few factors. Thanks, -Uplinger Took a look at the longest running 5 digit batch one that I’ve done and is still PV. Had 343 jobs and ran for 38 minutes on a machine that normally does the 4 digit batches in 1-5 minutes.
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.
  |
||
|
|
flynryan
Senior Cruncher United States Joined: Aug 15, 2006 Post Count: 235 Status: Offline Project Badges: 
|
It seems the transfers throughput is better with longer running work units. When the GPU's are finishing a job every minute, there are way more uploads and downloads than when the GPU's have to work for say 10 minutes per WU. It seems the number of file downloads and uploads are the problem with smaller work units.
|
||
|
|
lakotamm
Cruncher Joined: Jul 4, 2019 Post Count: 5 Status: Offline Project Badges:
|
Thanks! So I am not alone. So if I understand correctly, I would need much more CPU power for a task to keep the GPU busy all the time. (Or maybe it is even impossible?).
|
||
|
|
|