| 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 - Feb 27 2021 [ Issues Thread ] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 162
|
|
| Author |
|
|
Vester
Senior Cruncher USA Joined: Nov 18, 2004 Post Count: 325 Status: Offline Project Badges: 
|
Four PV on AMD Radeon HD 7990.
---------------------------------------- |
||
|
|
uplinger
Former World Community Grid Tech Joined: May 23, 2005 Post Count: 3952 Status: Offline Project Badges: 
|
Hi uplinger, is each step in the GPU WUs comparable to an "old" OPN CPU WU or are these steps smaller? thanks! They are equivalent to the CPU work units. They solve the same problem just a different way. But for each CPU work unit, they would run multiple ligands in a work unit (sometimes less if they were difficult). For GPU they run lots of ligands in a single work unit due to the speed of the calculations. Thanks, -Uplinger |
||
|
|
Dark Angel
Veteran Cruncher Australia Joined: Nov 11, 2005 Post Count: 721 Status: Offline Project Badges: 
|
So the credit return will be corrected for the production units? 2.5 for the same work done done by a CPU unit that gets 65 is a little off.
---------------------------------------- Currently being moderated under false pretences |
||
|
|
SOS_Ready
Cruncher Joined: Sep 23, 2008 Post Count: 19 Status: Offline Project Badges: 
|
I recently added 2 Nvidia GTX TITAN cards to my computer (AMD Opteron 6386SE) for this project. I wonder if someone can help me to answer three questions:
1. How to find the optimal number of tasks which can/should run in parallel on GPU and the number of CPUs supporting each GPU task in order to maximize the output of such process? I know how to manipulate these settings in the app_config.xml file but I could not find any empirical/practical formula to suggest the solution. Sure - trial and error will provide an answer too but maybe there is no need to reinvent the wheel. 2. Could activating SLI be beneficial for this kind of project? I know it enables parallel processing of one task on two identical GPUs but I doubt if this will be useful when we have each GPU already busy handling several tasks. 3. Is a double floating point precision on GPU a must or nice to have or "does not matter" for this project? Thanks! SOS_Ready |
||
|
|
wildhagen
Veteran Cruncher The Netherlands Joined: Jun 5, 2009 Post Count: 859 Status: Offline Project Badges:
|
I got 5 WU's yesterday on my GeForce GTX 970. All are Pending Validation now, and completed in about 8-12 minutes each.
|
||
|
|
Crystal Pellet
Veteran Cruncher Joined: May 21, 2008 Post Count: 1323 Status: Offline Project Badges: 
|
.... For GPU they run lots of ligands in a single work unit due to the speed of the calculations. Speed of calculation is relevant. The range in performance of GPU's is much greater than of CPU's.Thanks, -Uplinger I'm running atm a 60 jobs task on a R3 graphics apu and after 15 hours 89% done. |
||
|
|
William Albert
Cruncher Joined: Apr 5, 2020 Post Count: 39 Status: Recently Active Project Badges:
|
I recently added 2 Nvidia GTX TITAN cards to my computer (AMD Opteron 6386SE) for this project. I wonder if someone can help me to answer three questions: 1. How to find the optimal number of tasks which can/should run in parallel on GPU and the number of CPUs supporting each GPU task in order to maximize the output of such process? I know how to manipulate these settings in the app_config.xml file but I could not find any empirical/practical formula to suggest the solution. Sure - trial and error will provide an answer too but maybe there is no need to reinvent the wheel. BOINC will show the fractions of a CPU required to feed the GPU work. Given your old CPU, I would use 1 core per GPU as a rule of thumb. That being said, your GPUs are so much more powerful than your Opteron that trying to optimize CPU usage wouldn't be worth it. Crunching on the Opteron at all may even be counterproductive, as modern GPUs will throttle down their clock speeds if the internal case temperature is too high. 2. Could activating SLI be beneficial for this kind of project? I know it enables parallel processing of one task on two identical GPUs but I doubt if this will be useful when we have each GPU already busy handling several tasks. SLI is useless for compute workloads like BOINC. BOINC will automatically parallelize work across all of your GPUs simply by running multiple GPU jobs. |
||
|
|
bozz4science
Advanced Cruncher Germany Joined: May 3, 2020 Post Count: 104 Status: Offline Project Badges:
|
Any ETA as to when to expect new beta WUs to be distributed?
---------------------------------------- AMD Ryzen 3700X @ 4.0 GHz / GTX1660S Intel i5-4278U CPU @ 2.60GHz |
||
|
|
JohnDK
Advanced Cruncher Denmark Joined: Feb 17, 2010 Post Count: 77 Status: Offline Project Badges: 
|
So the credit return will be corrected for the production units? 2.5 for the same work done done by a CPU unit that gets 65 is a little off. Yss odd, I had 2 WUs in this 2nd batch each with 2.6 credit, the first batch gave normal credits.
Intel i7-6850K / 16GB / RTX 3090 / 2x RTX 3080 Ti / RTX 3070 Ti
AMD Ryzen 9 5950X / 32GB / RTX 2080 Ti |
||
|
|
Jorlin
Advanced Cruncher Deutschland Joined: Jan 22, 2020 Post Count: 90 Status: Offline Project Badges:
|
Four PV on GTX 660, about 45 minutes each.
---------------------------------------- |
||
|
|
|