The first one seems to take a long time on CUDA. Thinking maybe I should stick to PG on CUDA while the getting is good.
The first one seems to take a long time on CUDA. Thinking maybe I should stick to PG on CUDA while the getting is good.
Finnaly figured out my problems with Moo Wrapper and dual 5870s in CF. I ran the following.
"C:\Users\All Users\BOINC\projects\moowrap.net\dnetc518-win32-x86-stream.exe" -bench
Actually the above command is dependent where your Bonic project files are.
This came back and said core 3 was the best option for Moo Wrapper. I was running with the default which was core 0. I modified this using dnetXX --config. Didn't work. I then changed my Moo Wrapper preferences to use core 3. Now I've gone from 3600 to complete a work unit to 900 seconds, netting 12K.
Dan
Last edited by dan; 05-11-11 at 07:13 PM.
Will that result hold for single 5870s?
I ran it on 4 machines:
2x5870: 3 was the best by far
1x5870: 3 was just barely better than 0
1x4870: 0 was best
1x4870: 0 was best
-1 is supposed to auto-select the best. I don't trust it.
So I set up a home and a work preference, 3 and 0 respectively. Put the 5870s on home, and the 4870s on work.