| 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 |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I finally got a lot of work units to download most of which were CPU work units but a few of which were GPU work units
|
||
|
|
Digiloog
Cruncher Joined: Feb 12, 2021 Post Count: 4 Status: Offline Project Badges: 
|
The generated point are going through the roof with this stress test:
----------------------------------------[Edit 1 times, last edit by Digiloog at Apr 28, 2021 6:48:25 AM] |
||
|
|
Mumak
Senior Cruncher Joined: Dec 7, 2012 Post Count: 477 Status: Offline Project Badges: 
|
I'm still unable to get any GPU task on the Intel DG1 (Discrete Graphics 1). It seems the scheduler ignores it, which wouldn't be a big surprise as this is something not common, but it would be nice to solve this as we can (hopefully) expect more DGs from Intel.
----------------------------------------I reported this issue 2 times, here more details: https://www.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=655373  |
||
|
|
Chooka
Cruncher Australia Joined: Jan 25, 2017 Post Count: 48 Status: Offline Project Badges: 
|
I noticed that GPU usage is different for the newer batches. As soon as batches above 13000 started, the GPU usage of my GTX-1660S was not at 100% anymore. Please note that I'm running 4 WUs concurrently via app_config.xml: <app_config> <app> <name>opng</name> <fraction_done_exact/> <gpu_versions> <gpu_usage>0.25</gpu_usage> <cpu_usage>0.25</cpu_usage> </gpu_versions> </app> </app_config> See here for a larger image. Why is the GPU usage not at 100% anymore all the time? How do those batches differ? I have some Radeon VII's and I'm still struggling to work out how many wu's to run concurrently. GPU usage is still up and down like a yo yo running 3 wu's concurrently. I do like the low power usage of these work units though!   |
||
|
|
erich56
Senior Cruncher Austria Joined: Feb 24, 2007 Post Count: 295 Status: Offline Project Badges: 
|
That being said, the performance tab of Windows Task Manager should show reasonably-accurate disk throughput if you're just looking for a point-in-time snapshot. ah, I had forgotten about the Task Manager. So I looked at it now. Interestingly enough, it shows a write rate of about 7MB/s, with 3 of these current WCG tasks running. No idea why yesterday someone mentioned about 40MB/s. |
||
|
|
Chooka
Cruncher Australia Joined: Jan 25, 2017 Post Count: 48 Status: Offline Project Badges:
|
No word yet on which GPU's are performing better?
----------------------------------------I have a feeling from the limited numbers I've seen that NVIDIA is outperforming AMD GPU's, especially the 30xx series. |
||
|
|
DrMason
Senior Cruncher Joined: Mar 16, 2007 Post Count: 153 Status: Offline Project Badges: 
|
Today was pretty smooth. Congrats and thanks to uplinger and the team for getting those settings fixed up so that this is possible!
---------------------------------------- |
||
|
|
Andrew80431
Cruncher Joined: Nov 25, 2005 Post Count: 36 Status: Offline Project Badges: 
|
this graphic looks interesting - what's the name of this tool? This tool is named Grafana. Behind it is an InfluxDB that is being fed by Telegraf, which in term gets its data from the nvidia-smi command line tool.  |
||
|
|
spRocket
Senior Cruncher Joined: Mar 25, 2020 Post Count: 277 Status: Offline Project Badges:
|
No word yet on which GPU's are performing better? I have a feeling from the limited numbers I've seen that NVIDIA is outperforming AMD GPU's, especially the 30xx series. A complicating factor is which generations of GPUs. Comparing a GT 7xx to an RX 6xxx would be apples and oranges. At any rate, that old GTX 960 of mine has been handling things well. From last day's running, my main cruncher got about 252,000 BOINC credits, and about 240,000 of that was from the GPU. I went and checked Micro Center, and, wow, good luck getting a GPU that isn't either a) low-profile fanless and old, b) a Maxwell-generation Quadro that has far fewer cores than my GTX 960 and costs twice as much, or c) $2000+. Guess I won't be shopping for a GPU anytime soon. |
||
|
|
Azmodes
Cruncher Joined: Apr 4, 2017 Post Count: 3 Status: Offline Project Badges: 
|
thanks uplinger, any comment about what's going on with the low GPU utilization (lots of GPU idle time) of the 5-digit batches? you had mentioned that you though they should run fast. I even confirmed that the process is constantly comming on and off the GPU. you can catch times running nvidia-smi where it shows the wcg application isnt even running on the GPU, while BOINC shows it running. and it'll constantly pop in and out. this is much different than all the tasks before, where even if the sub jobs were starting and stopping, nvidia-smi still recognized that the application was running on the GPU. I'm seeing the exact same thing. GPU utilization is way down, CPU time ends up only being a quarter of the task (whereas it was about 100% before) and the processes keep showing up and vanishing again in nvidia-smi. Unsurprisingly runtimes appear to be longer. Anyone? This is kind of weird, I'm only seeing this behaviour on one of my hosts, which makes me think it's something on my end after all. Runtimes are getting abysmal, CPU load dropping to 20% or below. I can't seem to figure this out, particularly since the host runs fine on all other projects. CPU: 1950X GPUs: 2070S, 2060 Driver: 450.80 OS: Ubuntu 20.04.2 LTS [5.8.0-50-generic|libc 2.31 (Ubuntu GLIBC 2.31-0ubuntu9.3)] [the only difference that I can see right now is that this one's glibc version is 9.3, whereas all my other computers are 9.2] |
||
|
|
|