Reached daily quota of 1024 tasks
9 Sep 2018 22:45:58 UTC
Topic 216307
(moderation:
Is there a way to bypass the 1024 daily task quota? WU times improved recently and I think my 3 Vega rig is now getting through 1024 tasks in less than 24 hrs. The last few days I've had the problem below. Any ideas?
9/9/2018 6:37:55 PM | Einstein@Home | Requesting new tasks for AMD/ATI GPU
9/9/2018 6:37:56 PM | Einstein@Home | Scheduler request completed: got 0 new tasks
9/9/2018 6:37:56 PM | Einstein@Home | No work sent
9/9/2018 6:37:56 PM | Einstein@Home | (reached daily quota of 1024 tasks)
The daily quota is controlled
)
The daily quota is controlled by both the number of GPUs and the number of CPU cores that BOINC detects. You can get BOINC to assume an increase in the number of CPU cores in your machine (even though it isn't 'real') just by setting the parameter 'ncpus'.
You have to set up a cc_config.xml file and include a line in the <options> section saying
<ncpus>N</ncpus>where N is perhaps double the true number of cores that you have. That should give you a significantly increased quota.
I haven't actually tried this but I know it works because of this message where Gavin solved a similar problem some time ago. I don't image you even need to stop and restart BOINC. If you create cc_config and drop it into place, you should just be able to click the "reread config files" option in Boinc Manager and then try updating the project to tell it about all your new cores :-).
Let us know how it goes -).
Cheers,
Gary.
Works great, thanks Gary! I
)
Works great, thanks Gary! I switched my core count to 16 (from 8) and my quota went up to 1280 (from 1024), which should be enough to get me through the day.
Gavin had the same issue, also with a Vega 64. It does seem like a bump to the GPU daily quota is in order as were already hitting the limit and the GPUs will only get faster...
I'm glad the solution worked
)
I'm glad the solution worked for you.
The problem arose last time for Gavin because of a change in the nature of the tasks at that time. The same thing has happened again now.
Recently, we were running tasks based on LATeah1019L.dat (and earlier) which had been taking what I would call the 'normal' amount of time. There didn't seem to be quota problems with those. When that lot finished and we had the outage for about a day, the new tasks for the replacement data (LATeah0104M.dat and now LATeah0104N.dat) have been taking quite a lot less time to complete - hence the pressure again on daily quotas for the very high end GPUs.
If this current fast running set is just a temporary arrangement until the 'normal' tasks are ready to run again, maybe nothing needs to change. I'm sure the Admins would be aware of this and will come up with a solution if the fast running tasks are to continue.
I guess one of the difficulties is that there may be a huge difference between a current 'top-of-the-range' model and anything somewhat older or mid to low end. If you set a quota high enough for the top end, you open up the potential for an older, misbehaving unit to trash a huge number of tasks before exhausting the supply. I guess the ideal would be to be able to broadly categorise GPUs so that only the modern, high end units got a higher quota.
Sounds simple, but probably a lot of coding effort to implement and perhaps extra load on servers to apply.
Cheers,
Gary.
I think you could also enable
)
I think you could also enable the HD 530 GPU which is integrated in the Core-i CPU and increase the quota by another 256. Boinc would need to recognize the OpenCL capability of the HD 530. No need to actually allow (in the preferences of your account) tasks for the Intel-GPU.