This is the most scherzando posting in a while....

Give it some time, it will re-calculate it

Cheers

Maybe patch Tuesday flunked your display drivers [Mine took 295Mb on the W7-64 system]. Can't imagine why the client would suddenly give up, and 6.10.xx definitely less detection capable. Of course, you've done a boot and considered posting the top 35 lines of the startup log to inform us how your client is configured.

How much time do you think it should need?

Old clients recalculated DCF and TTC every task that is completed, aggressively up when longer, slowly down when faster. Since <dont_use_dcf/> is put in place [a WCG server instruction only working for the WCG project tasks], clients understanding that instruction just follow the mean flops that's transmitted in task headers of new work assignments. This projecting is at the app level, so theoretically the CPU tasks would not be affected by the GPU Tasks actual runtimes, but since the tasks come from the same pool, the FLOPS used to project the time is likely equal.

Of course, if the client benchmark has not run [I've disabled it with <skip_cpu_benchmark>], then fast or slow CPU speeds, the projections are pretty constant, at least, on my CEP2/HCC1 mix octo, the HCC just loiters between 3:37 and 3:41. The long nor the shorts affect the TTC of still to start jobs, fractions of a minute for CPU tasks, whenever new work arrives [never sat there and watched it happen, so it's big thumb theory].