This problem pertains to CUDA users.

The problem I'm having is that if I set my preferences to allow seti-enchanced & astropulse, two scenarios happen frequently:

1) I get ALL seti-enchanced-CUDA work units. Since my queue is now full, it won't even check for other work units that would put my CPUs to work.

2)I get ALL astropulse or simple-seti work units. Again, now my queue is full, and won't look for CUDA work units that my GPU could use.

I believe that the queue should be seperated for CPU work units and GPU work units. Or am I missing something?

The upcoming BOINC 6.6 client has a separated CPU and GPU scheduler. This will easier fetch work for both than the present 6.4 clients do, as they are hampered by only having one CPU scheduler that's supposed to cater for both.

---