Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 10
|
![]() |
Author |
|
squaregoldfish
Cruncher Joined: Dec 18, 2005 Post Count: 8 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I'd like to run OpenPandemics on my GPU and Africa Rainfall Project on my CPU. Is there a way to configure WCG to only give me GPU work units for OPN? I assume that if I sign up for both OPN and ARP, I'll be getting CPU units for both projects and GPU OPN units when they're available.
|
||
|
zdnko
Senior Cruncher Joined: Dec 1, 2005 Post Count: 225 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
sam6861 suggest to keep only ARP enabled and select "If there is no work available for the project(s) I have selected above, please send me work from another project." so as there is no other GPU work it will still grab GPU tasks from OpenPandemics.
----------------------------------------https://www.worldcommunitygrid.org/forums/wcg...d,43290_offset,280#654262 It's a workaround (some time you can get wu from other cpu project), would be better have a option to do that. [Edit 2 times, last edit by zdnko at Apr 7, 2021 7:41:53 AM] |
||
|
squaregoldfish
Cruncher Joined: Dec 18, 2005 Post Count: 8 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Thanks, I'll try that and see how it goes.
|
||
|
squaregoldfish
Cruncher Joined: Dec 18, 2005 Post Count: 8 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Didn't work, sadly.
I think the problem is that I restrict ARP to only run on half my cores, so WCG thought it could fill up the other cores with WUs from the other projects. |
||
|
nanoprobe
Master Cruncher Classified Joined: Aug 29, 2008 Post Count: 2998 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Didn't work, sadly. I think the problem is that I restrict ARP to only run on half my cores, so WCG thought it could fill up the other cores with WUs from the other projects. The easiest way to accomplish what you want is to use 2 separate profiles. One for GPU tasks that has send work for CPU set to no and another for the CPU tasks you want with send work for CPU set to yes. Use that profile to fill your cache with CPU tasks and then switch to the other and wait for GPU tasks while running the CPU tasks. When the CPU tasks run low switch profiles and load up again. There are more complicated ways like installing and running 2 instances of BOINC or using a VM. I've done the 2 profile thing before and it works but takes some baby sitting.
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.
![]() ![]() |
||
|
Jorlin
Advanced Cruncher Deutschland Joined: Jan 22, 2020 Post Count: 89 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I just do a compromise and limit the OPN1 task to max_concurrent 1
----------------------------------------![]() |
||
|
cehunt
Senior Cruncher CANADA Joined: Oct 10, 2011 Post Count: 172 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Hi:
I have a similar issue. I have a laptop that has a video card that cannot run Openpademics GPU due to its older video card. So I thought I would set up two device profiles. One for the laptop and another one for the desktop. 1. I have tried to set up a device profile called Home on the laptop to run only the CPU version of Openpademics. 2. On the desktop which has a newer CPU and GPU, I have set up another device profile called Work to run both versions of Openpandemics. When I viewed the results on EVENTS, only the Work profile shows up and the GPU Openpandemics work units are showing up on the laptop. Does anybody out there have a better idea? Clive |
||
|
Speedy51
Veteran Cruncher New Zealand Joined: Nov 4, 2005 Post Count: 1290 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Just wanting to make sure I have things set correctly. When you don't receive GPU work do you get the following messages? TIA
----------------------------------------9/04/2021 3:05:49 PM | World Community Grid | Sending scheduler request: To fetch work. EdIT = added answer to my own question I answered my own question and the answer is yes ![]() [Edit 1 times, last edit by Speedy51 at Apr 9, 2021 4:30:35 AM] |
||
|
sam6861
Advanced Cruncher Joined: Mar 31, 2020 Post Count: 107 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
On Profile settings,
- Turn off "OpenPandemics" - Turn on "If there is no work available for the project(s) I have selected above, please send me work from another project." - Turn on both CPU and graphics card. - Set project limits to unlimited to one of the other selected active project (MIP1 MCM1 ARP1). - Set project limit for OpenPandemics to 10 or higher or unlimited. If this is set to 1, will limit OPNG to only 1 task at a time, which is bad if computer have more then 1 GPU. This similar configuration easily works on Milkyway at home to prioritize CPU to selected project but they don't have option to change project limit. This appears to work for my computer, the primary problem is too little amount of GPU work available. Just requires some waiting time I guess. Here is a small log of the few times it got GPU work. 08-Apr-2021 03:16:40 [World Community Grid] Sending scheduler request: To fetch work. 08-Apr-2021 03:16:40 [World Community Grid] Requesting new tasks for AMD/ATI GPU 08-Apr-2021 03:16:41 [World Community Grid] Scheduler request completed: got 3 new tasks 08-Apr-2021 03:16:41 [World Community Grid] No tasks are available for the applications you have selected 08-Apr-2021 03:16:41 [World Community Grid] No tasks are available for Africa Rainfall Project 08-Apr-2021 03:16:41 [World Community Grid] No tasks are available for Microbiome Immunity Project 08-Apr-2021 03:16:41 [World Community Grid] No tasks are available for Help Stop TB 08-Apr-2021 03:16:41 [World Community Grid] No tasks are available for Smash Childhood Cancer 08-Apr-2021 03:16:41 [World Community Grid] Your preferences allow tasks from applications other than those selected 08-Apr-2021 03:16:41 [World Community Grid] Sending tasks from other applications 08-Apr-2021 03:16:41 [World Community Grid] Project requested delay of 121 seconds 08-Apr-2021 03:16:46 [World Community Grid] Starting task OPNG_0000497_00343_1 |
||
|
poppinfresh99
Cruncher Joined: Feb 29, 2020 Post Count: 49 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() |
On Profile settings, - Turn off "OpenPandemics" - Turn on "If there is no work available for the project(s) I have selected above, please send me work from another project." - Turn on both CPU and graphics card. - Set project limits to unlimited to one of the other selected active project (MIP1 MCM1 ARP1). - Set project limit for OpenPandemics to 10 or higher or unlimited. If this is set to 1, will limit OPNG to only 1 task at a time, which is bad if computer have more then 1 GPU. The above trick for sure does not work on my Intel GPUs. During the OPNG stress test, I used the above settings and had auto clickers clicking the Update button in BOINC (every 300 seconds). I never got any tasks from this. However, I could easily get tasks by turning on OpenPandemics for between a minute and a couple hours until I got a bunch of OPNG tasks, then I'd turn off OpenPandemics. I did this successfully multiple times at various times of the day on a couple different Intel GPUs. I thought the "If there is no work available for the project(s)" trick could still work because my Event Log would say "Requesting new tasks for Intel GPU", but, when I deactivated both OpenPandemics and turned off "If there is no work available for the project(s)", the Event Log would still say "Requesting new tasks for Intel GPU". Here are some hypotheses I have... - The "If there is no work available for the project(s)" trick doesn't work with any GPUs. I read once in these forums that BOINC client needs to update a few times to get settings from WCG causing the appearance of the trick working when BOINC client is actually using old settings. Or, WCG may have changed something on their end. - Intel GPUs are treated differently than AMD GPUs by OPNG. - I am a statistical outlier!! |
||
|
|
![]() |