I have tons of CUDA WU's, but ran out of WU's for the CPU over a week ago. Even though BOINCMGR requests both CPU and GPU, I'm only getting GPU WU's.

I think the opposite happened a while back when the pipes were clogged...

Seems there should be more intelligence behind this...

Any suggestions?

---

Disable GPU work fetch for the venue of that computer in your SETI@Home preferences, now your BOINC client should request work only for CPU (from the 2nd request after that change), once you have enough work for your CPU, enable GPU work fetch again.

---

The server's scheduler will fill the fastest device first, lower your cache settings until Boinc no longer asks for GPU work,

Claggy

The server's scheduler will fill the fastest device first, lower your cache settings until Boinc no longer asks for GPU work,

Claggy

Or allow work from the 2 other projects like Einstein and Rosetta, in which
you've partissipated before.

This isn't a 'race' and SETI does have the most partissipants compaired to
other projects, that'll garantee you'll never run out of CPU, also GPU work.

My ATI GPU(s) system runs SETI & Einsteins Bibainy Pulsar Search, each on
'half' HD5870s GPUs.

Both can be running on the same Single Precision GPUs.

---

Thanks for all the help. Looks like I'm getting something through the clogs now.

I suspect this condition has more to do with a policy of 'give-what-you-get'. If CPU WU's are being reported, deliver replacement CPU WU's. Same for GPU. Once I had run out of CPU WU's (due to a shortage?), I was in a vicious cycle.

Just a theory...

---