Moo WU taking an Hour per on 5870 and 5990

[Mike029]

Hey Mike,

I sounds like Moo is CPU starved. Have you tried editing the POEM app_info file <avg_ncpus></avg_ncpus> and <max_ncpus></max_ncpus> ? Try 0.5 and see if that helps.

Did not work. Can't understand why it's choking out when I assign 2 cores to Moo? I assign 4 to Poem. I even reduced the priority of Poem to below normal and raised Moo to high. No go. I stop using Prolasso, same results Moo is choked out.

I've also noticed that Poem takes 10 minutes longer to run when Moo is running vs Milk. I'll continue to play with this.

I think I figured this out.
10/11/2012 10:15:57 AM | | NVIDIA GPU 0: GeForce GTX 560 Ti (driver version 306.23, CUDA version 5.0, compute capability 2.1, 1024MB, 8381357MB available, 1382 GFLOPS peak)
10/11/2012 10:15:57 AM | | ATI GPU 0: Cypress (CAL version 1.4.1741, 1024MB, 991MB available, 5760 GFLOPS peak)
10/11/2012 10:15:57 AM | | ATI GPU 1: Cypress (CAL version 1.4.1741, 1024MB, 991MB available, 4928 GFLOPS peak)
10/11/2012 10:15:57 AM | | ATI GPU 2: Cypress (CAL version 1.4.1741, 1024MB, 991MB available, 4928 GFLOPS peak)


It seems that my AMD cards are running as devices 0,1,2. I have the Cuda card in the first slot closest to the cpu. Boinc Afterburner has labled the Cuda card as GPU4. Funny part is that it does not impact my Milk wu completion times at all. Just my Moo.
Hmm, how to fix this? If I can fix this I'll fix my problem.