[solved] PARDISO doesn't recognize HTT threads

[Sören_Plönnigs]

Hi,

I want to do some benchmarking on pardiso. I have the problem that pardiso doesn't recognize HTT Threads.

I now that there won't be a performance gain, but I need this comparison for my bachelor thesis.

Our System has 8 Cores and 16 Threads.

Neither specifying iparm[2]=16 nor setting OMP_NUM_THREADS=16 does the job.

HTT is enabled in BIOS as the following output shows:

OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 10 maps to package 0 core 0 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 1 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 14 maps to package 0 core 1 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 1 core 0 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 15 maps to package 1 core 0 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 1 core 1 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 11 maps to package 1 core 1 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 2 core 0 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 8 maps to package 2 core 0 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 2 core 1 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 12 maps to package 2 core 1 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 3 core 0 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 13 maps to package 3 core 0 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 3 core 1 thread 0

OMP: Info #171: KMP_AFFINITY: OS proc 9 maps to package 3 core 1 thread 1

OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0

The same problem applies to ISS and DSS.

Does anybody know what's the problem?

Regards,

Soeren

[Konstantin_A_Intel · Accepted solution]

Hi Soeren,

MKL dynamically changes the number of threads to the most optimal value. To disable this, please usemkl_set_dynamic(0) call.

With best regards,

Konstantin

[Gennady_F_Intel]

Soeren,

just two my cents -yes, thesame problem applies to ISS and DSS and to all others MKL domains. Using MKL_DYNAMIC Environment variable and/or mkl_set_dynamic() service function will enable MKL to dynamically change the number of threads. By the default,

MKL_DYNAMIC is TRUE and whenMKL tries to use what it considers the best number ofthreads, up to the maximum number you specify.

see more details into User's Guide.

--Gennady