| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Hardware Chat Room
Thread: Simultaneously running multiple clients on different GPUs? Is it possible? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges: 
|
Hi,
----------------------------------------as a hobby/test, I acquired some old second hand GPUs which I'm planning to install on a machine with rtx2070 already in it . These GPUs are low performance cards (GT710, GT1030, GTX 1050) . I would like to run these cards on different settings than 2070. Is it possible to simultaneously run multiple clients on on a single workstation, so for example Client A would use RTX2070 and Client B would run lower performance GPUs? E: Running win10 Pro on test machine ---------------------------------------- [Edit 1 times, last edit by biini at May 21, 2021 12:50:48 PM] |
||
|
|
Acibant
Advanced Cruncher USA Joined: Apr 15, 2020 Post Count: 126 Status: Offline Project Badges:
|
You can use BOINC advanced client configuration settings such as the <exclude_gpu> option as described here. I'm not sure if you can install two BOINC clients pointing at WCG to then assign different device profiles with each excluding different GPUs, though. At the least they'd have to be installed into different directories.
----------------------------------------WCG also doesn't have much GPU work so another option would be to exclude one or more GPUs as described above for them to be ignored by WCG and install Folding@Home. Then use advanced slots configuration to use that project's settings to apply the client to particular GPUs (remember to cut down the CPU slots to just the number of GPUs used to leave threads free for WCG). You could then have F@H set to run only when the computer is idle and WCG to always run, for example.  |
||
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges:
|
Hi,
----------------------------------------thank you for your reply, I'll try exclude the GPUs. I have a test machine running with two GPUs now, but it seems that one client cannot handle two GPUs of the same manufacturer simultanously 21/05/2021 22:32:22 | | CUDA: NVIDIA GPU 0: GeForce GTX 1050 Ti (driver version 456.71, CUDA version 11.1, compute capability 6.1, 4096MB, 3376MB available, 2138 GFLOPS peak) ---------------------------------------- [Edit 1 times, last edit by biini at May 21, 2021 7:54:07 PM] |
||
|
|
ca05065
Senior Cruncher Joined: Dec 4, 2007 Post Count: 328 Status: Offline Project Badges: 
|
In the cc_config.xml file there is an option:
<use_all_gpus>0|1</use_all_gpus> If 1, use all GPUs (otherwise only the most capable ones are used). Requires a client restart. |
||
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges:
|
Hi,
----------------------------------------Yes, use_all_gpus is set to 1 |
||
|
|
captainjack
Advanced Cruncher Joined: Apr 14, 2008 Post Count: 144 Status: Offline Project Badges: 
|
Running multiple BOINC clients on one PC is possible. Here is a guide for setting it up on Windows 10.
https://www.overclock.net/threads/guide-setti...-boinc-instances.1628924/ |
||
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges:
|
It seems that I cannot get the gt1030 working with gtx1050.
----------------------------------------No matter what I do, it's "ignored by config" . Any ideas? 23/05/2021 01:35:12 | | CUDA: NVIDIA GPU 0: NVIDIA GeForce GTX 1050 Ti (driver version 466.47, CUDA version 11.3, compute capability 6.1, 4096MB, 3376MB available, 2138 GFLOPS peak) 23/05/2021 01:35:12 | | CUDA: NVIDIA GPU 1 (not used): NVIDIA GeForce GT 1030 (driver version 466.47, CUDA version 11.3, compute capability 6.1, 2048MB, 1661MB available, 1127 GFLOPS peak) 23/05/2021 01:35:12 | | OpenCL: NVIDIA GPU 0: NVIDIA GeForce GTX 1050 Ti (driver version 466.47, device version OpenCL 3.0 CUDA, 4096MB, 3376MB available, 2138 GFLOPS peak) 23/05/2021 01:35:12 | | OpenCL: NVIDIA GPU 1 (ignored by config): NVIDIA GeForce GT 1030 (driver version 466.47, device version OpenCL 3.0 CUDA, 2048MB, 1661MB available, 1127 GFLOPS peak) |
||
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges:
|
Turns out I had a stick of invalid memory. Removed the defect memory, reinstalled win10 and client and now it seems to working as desired. Just have to wait and see now if I can get any work for these cards :-)
---------------------------------------- |
||
|
|
biini
Senior Cruncher Finland Joined: Jan 25, 2007 Post Count: 334 Status: Offline Project Badges:
|
Is it possible to change the instance name that is running other card? Not a biggie, but now I have the same computer name twice in device manager and results status pages, which can get confusing after adding more GPUs. I was hoping to name them like MyDesktop_1050, MyDesktop_1030 etc
---------------------------------------- |
||
|
|
captainjack
Advanced Cruncher Joined: Apr 14, 2008 Post Count: 144 Status: Offline Project Badges:
|
|
||
|
|
|