I'm back to the occasional bundles of GPU tasks. It's working well enough that I don't feel a need to worry about it, and with scarce GPU tasks instead of a continuous supply, I've renamed my app_config.xml out of the way and gone back to 15 CPU tasks.

As an aside, one thing about going back to the small batches of GPU tasks is that it levels the playing field at least somewhat. Even a modest GPU can crank out plenty of work. No need to break the bank for an RTX 3080, assuming you can even get one.



This is what I was using before going back to one GPU task. This is on a Ryzen 7, 14 CPU tasks allocated by Computing Preferences, and a GTX 960 running two tasks.

<app_config>
   <app>
      <name>opng</name>
      <max_concurrent>2</max_concurrent>
      <gpu_versions>
          <gpu_usage>.5</gpu_usage>
          <cpu_usage>.25</cpu_usage>
      </gpu_versions>
   </app>
</app_config>

The max_concurrent setting might not be necessary, but I use it as insurance.

Well, no GPU tasks at all since the stress test ended. Shutting down my main GPU cruncher.

Yup, information died with the end of the stress test.
And so did all GPU tasks it seems. It was hard to get GPU tasks before the stress test, but now it's totally impossible. I have tried since the stress test ended, but with no luck at all.

So, my second GPU cruncher goes offline now too.

I'm finding that if I just let it randomly grab GPU work units when they're available, I'm generally getting enough to more than double my daily points. So, really, I can't complain, especially since that system is my primary cruncher. Sure, the numbers were downright insane during the stress test (especially for anyone who ran multiple tasks), but it's going to be a challenge to keep the pipe full for any length of time.

Sometimes I might poke the server and rarely dislodge a few OPNGs, but BOINC is doing a good job on its own most of the time.

Hmm, why do we still have a 3-day deadline on the new non stress test WU's?

Just one example: https://www.worldcommunitygrid.org/ms/device/...s.do?workunitId=660785640