Thread Affinity

toddwz — Thu, 17 Oct 2024 13:08:35 GMT

I have 2 questions when reading https://www.intel.com/content/www/us/en/docs/dpcpp-cpp-compiler/developer-guide-reference/2024-2/thread-affinity-interface.html.

1 It says

Binding OpenMP threads to particular packages and cores will often result in a performance gain on systems with Intel processors with Intel® Hyper-Threading Technology (Intel® HT Technology) enabled; however, it is usually not beneficial to bind each OpenMP thread to a particular thread context on a specific core.

Is binding beneficial or not?

2 Specifying KMP_AFFINITY=verbose,granularity=core,compact, it shows

KMP_AFFINITY: OS proc 0 maps to socket 0 core 0 thread 0 
KMP_AFFINITY: OS proc 4 maps to socket 0 core 0 thread 1 
KMP_AFFINITY: OS proc 2 maps to socket 0 core 1 thread 0 
KMP_AFFINITY: OS proc 6 maps to socket 0 core 1 thread 1 
KMP_AFFINITY: OS proc 1 maps to socket 3 core 0 thread 0 
KMP_AFFINITY: OS proc 5 maps to socket 3 core 0 thread 1 
KMP_AFFINITY: OS proc 3 maps to socket 3 core 1 thread 0 
KMP_AFFINITY: OS proc 7 maps to socket 3 core 1 thread 1

Should proc 1 map to socket 0 core 1 thread 0?

Thanks.

Re: Thread Affinity

Alex_Y_Intel — Fri, 18 Oct 2024 17:29:46 GMT

Escalated your issue for further investigation.

topic Thread Affinity in Intel® oneAPI DPC++/C++ Compiler

Thread Affinity

Re: Thread Affinity