Bitcoin Forum

Bitcoin => Mining => Topic started by: callmeivan on June 03, 2011, 10:28:51 AM



Title: CPU bottleneck? (Q6600 Quad @ 2,8GHZ / ATI 5970 @ 925 MHZ)
Post by: callmeivan on June 03, 2011, 10:28:51 AM
Hi,

i believe my CPU may be a bottleneck in my configuration. I get around 2x380 MH/s, CPU usage shows exactly 50% so i guess both GPU Cores use one full CPU core without taking advantage of the other two remaining ones.

Is there a flag to enable multi threading on all cpu cores so i can leverage all of my GPU power or am i interpreting things wrong?

Thanks :)


Title: Re: CPU bottleneck? (Q6600 Quad @ 2,8GHZ / ATI 5970 @ 925 MHZ)
Post by: padrino on June 03, 2011, 10:32:41 AM
I tink you are dealing with a bug with some combinations of ATI drivers and miners (poclbm?). If using the GPU the CPU should be almost 0% but I've seen the same setup with mine when I OC, does not impact the has rate unless I pound my CPUs with something else, I also CPU mind but supply virtual cores - 4 to give my miners room.


Title: Re: CPU bottleneck? (Q6600 Quad @ 2,8GHZ / ATI 5970 @ 925 MHZ)
Post by: callmeivan on June 03, 2011, 10:38:39 AM
Yeah i use poclbm,

thank you padrino - was kinda confused why it would consume so much power in comparison to my osx machine but i thought my 5970 ist just so much more powerful than my 4870 in the osx machine :D, i am gonna try different driver versions.


Title: Re: CPU bottleneck? (Q6600 Quad @ 2,8GHZ / ATI 5970 @ 925 MHZ)
Post by: Genrobo on June 03, 2011, 12:52:35 PM
People use Sempron 140's in 4 card mining rigs.
I doubt it's bottlenecking you.


Title: Re: CPU bottleneck? (Q6600 Quad @ 2,8GHZ / ATI 5970 @ 925 MHZ)
Post by: Wepp on June 03, 2011, 01:56:40 PM
Try setting the processor affinity of both the miner processes to use the same core by right-clicking on them in the Windows Task Manager and setting them to the same core.

You'll see your CPU usage drop to a solid 25%, but the mining performance will remain the same. Actually, the computer should be a little more responsive.