Dear Members, we are here to assist each other. Please abide by our forum rules. Thanks.

Tedi

Ho blano1,

A rather vexing problem. You say *was*, implying you got some from elsewhere in the meantime, which at least infers that some of the client control counters are not corrupted. On the website, device profiles, there's an option to specify if only wanting GPU tasks. "Allow research to run on my CPU? Yes" (Default). Plz check if it really is.

Whilst, I do not know if 7.0.33 is actually smart enough to know that the device profile would have a "No" and thus the client responding with [don't need].

Let us know.

If you go into the project tab, select each project, one at the time and hit the properties button on left, you get supplemental information on scheduling and fetching for both GPU and CPU. If the numbers are zero or negative, than indeed WCG is not need [entitled] to CPU time. Mind you, they're still busying at the developers of the issue of GPU tasks causing CPU's to idle. The rather brute force trick, as was suggested earlier in this thread, is to briefly suspend all your other projects so WCG is truly the sole one that has work fetch rights. Then hit update to see what happens.

... 2 more computers with same issue? Amazing, if so.

Beyond me. Each has on the Advanced menu a "Do network communication". That resets any network deferral counters. 7.0.33 is pretty good at telling why work is not being supplied [a good dozen messages], so if e.g. work fetch is suspended, it would tell so i.e. not considering this. No rhyme or reason... not anything to go on.