topic Achieving peak bandwidth on multi-socket systems in OpenCL* for CPU

Achieving peak bandwidth on multi-socket systems

James_R_ — Wed, 26 Mar 2014 04:16:09 GMT

Let's say each CPU socket has 43 GB/s of bandwidth through its four memory channels. Let's say I have a dual socket system. A reduction operation should achieve performance of 86 GB/s, but it doesn't. It will still only achieve up to 43 GB/s. Why is that and is there anything in Intel's OpenCL implementation for CPUs that can fix that?

How could I fix that outside of OpenCL?

Dear James,

Arik_N_Intel — Wed, 26 Mar 2014 13:31:05 GMT

Dear James,

Can you please check if the following helps?

http://software.intel.com/forums/topic/497429

Thanks,

Arik

Quote:Arik Narkis (Intel)

James_R_ — Fri, 28 Mar 2014 05:03:24 GMT

Arik Narkis (Intel) wrote:

Dear James,

Can you please check if the following helps?

http://software.intel.com/forums/topic/497429

Thanks,

Arik

Arik,

This method improves bandwidth performance substantially (by >1.8x). It actually achieves 90+% of the platform bandwidth for my code rather than the ~50% of peak bandwidth I had the other day. I'm a bit surprised it actually worked. I'd like to test it on a four or eight socket system now, but I'll have to find one.

Knowing this now makes life more difficult for OpenCL developers with bandwidth-bound kernels on multi-socket nodes. Thanks a lot, Arik!

-James