Author |
Message |
|
Trying to get some beta GPU WU's and this is what I've tried:
- Created a WORK device profile and assigned it to the GPU host with an NVidia GTX 1060 and AMD R9 280x cards (are these cards too old? Does WCG require drivers no older than 3 months?)
- Set WORK profile to accept beta work.
- Set WORk profile to only accept COVID19 WU's.
- "Please opt me in to new projects as they become available" is checked.
- Allow CPU work on this profile = NO
- Allow GPU work while computer is in use = YES
- Allow work for AMD, NVIDIA, INTEL all = YES.
- Connect to network about every 0.01 days
- Leave applications in memory while suspended? = NO
- RSS share for WORK profile is set to 250.
- The GPU computing host's other project for GPU work is set to RSS share 0 and is only sending just-in-time single WU's every 4-7 mins.
- BOINC client is set to store up to 0.01 additional hours of work (which is supposed to minimize work request time interval).
- The "allow work while computer is in use", RAM and hard drive usage are set to 98% for 24/7 use BOINC client and plenty of hard drive space left.
I'm going to change these 2 options to see if they have an effect:
- Write to disk at most every: 30.0 sec -> 1.0 sec
- Leave applications in memory while suspended? No -> YES
The only other thing I can think of doing is just stop running any GPU work on that machine but that goes against all my instincts.
An idle card is an evil card!
Also, of course, I'm just not being lucky enough to grab the few WU's in beta that are available... |
|
|
|
Trying to get some beta GPU WU's and this is what I've tried:
- Created a WORK device profile and assigned it to the GPU host with an NVidia GTX 1060 and AMD R9 280x cards (are these cards too old? Does WCG require drivers no older than 3 months?)
- Set WORK profile to accept beta work.
- Set WORk profile to only accept COVID19 WU's.
- "Please opt me in to new projects as they become available" is checked.
- Allow CPU work on this profile = NO
- Allow GPU work while computer is in use = YES
- Allow work for AMD, NVIDIA, INTEL all = YES.
- Connect to network about every 0.01 days
- Leave applications in memory while suspended? = NO
- RSS share for WORK profile is set to 250.
- The GPU computing host's other project for GPU work is set to RSS share 0 and is only sending just-in-time single WU's every 4-7 mins.
- BOINC client is set to store up to 0.01 additional hours of work (which is supposed to minimize work request time interval).
- The "allow work while computer is in use", RAM and hard drive usage are set to 98% for 24/7 use BOINC client and plenty of hard drive space left.
I'm going to change these 2 options to see if they have an effect:
- Write to disk at most every: 30.0 sec -> 1.0 sec
- Leave applications in memory while suspended? No -> YES
The only other thing I can think of doing is just stop running any GPU work on that machine but that goes against all my instincts.
An idle card is an evil card!
Also, of course, I'm just not being lucky enough to grab the few WU's in beta that are available...
I will talk to my teammate who had to abort some as he didn't want them as to what he did to get them. I did essentially the same thing you did and am not getting any either. |
|
|
|
Project questions should be asked on that project's forum.
There is no work, what is to be expected. |
|
|
|
Project questions should be asked on that project's forum.
There is no work, what is to be expected.
That makes sense |
|
|
|
Project questions should be asked on that project's forum.
In my experience, people here are usually better at knowing the answer to getting rare WU's than the project support teams.
I'll see what they say.
There is no work, what is to be expected.
I noticed that after starting this thread while making the changes to the WORK profile.
I'd assumed there was work since seeing 98-114 hours in Bok's app for new WU's for the last 8 days. Someone could have gotten a 10+7 days worth of WU's in their cache or Bok's app is seeing a developers computer which runs WUProps (can't check ATM, Free-DC is offline https://isdown.me/stats.free-dc.org) |
|
|
|
OK fine, there is very little work. I've gotten I think two tasks that I've never seen run for 0.35 hours. But the instant flood of tasks that was expected.
The question was asked on the WCG forums. Can also see the batch release of tasks.
https://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,43206_offset,90#651952 |
|
|
|
AT&T have been down in my are since midnight last night unfortunately so Free-DC is down till they come back up. (And SETI-US as I host them too ) |
|
|
|
More Betas coming;
Posted at 18:06 UTC
Good afternoon,
We have received an updated pull request from the researcher to fix a bug that has been identified from the beta tests. We will need to run at least 2 more beta tests to confirm things are in better shape.
1. This will have the code changes from researchers
2. quick beta to remove the dlg file from being uploaded (this is just to make sure we don't break validation/assimilation and is quick)
Thanks,
-Uplinger |
|
|
|
Just posted
Hello everyone,
We are going to be starting a new beta test for OpenPandemics GPU. This will be for all 9 versions of the GPU application (Windows, Mac, Linux on AMD, Intel, NVIDIA). The version number of this application is 7.28. This test will include a change to the results sent back and a fix for an issue caught from earlier betas. We have also made some changes to the way maximum elapsed time is allowed per result to help prevent machines from reporting max time exceeded. I will be starting this with a few batches at first to make sure that things are operating properly, and then I will put more on the queue once I feel comfortable with how the beta is going. Because of this, I do not have an exact estimate of how many batches/work units will be sent out. If things are going well, you may see more work units released over the next few days.
Thanks again for your help,
-Uplinger |
|
|
|
My machines received 4 total WU's!
About 0.5 hours of work! |
|
|