Problem of multithreading mkl_?csrcsc

[Hao_W_]

Hey Guys

When I use mkl_?csrcsc to transpose my sparse matrix, I think the multithreading is supported internally: just like multithreading BLAS function cblas_dgemm(), call omp_set_num_threads() to set the number of threads before cblas_dgemm(). Unfortunately, no matter how many threads I set in omp_set_num_threads(), the performance of mkl_?csrcsc looks consistent.  I'm wondering how I can enable the multithreading for mkl_?csrcsc() functions? 

I use Intel compiler 13.0.1 on CentOS release 6.3, and my CPU is Intel(R) Xeon(R) CPU E5-2670.

Any suggestions are welcome.

- Hao

[Kenneth_C_Intel]

Hi, this is not one of our threaded functions so specifying to use more threads will have no effect. Are there any other questions that I can help you with?

Regards,

Kenneth

Intel Developer Support

[Gennady_F_Intel]

Hao, we are thinking about that and the problem is escalated to implement this feature into one of the future releases / updates. We will let you know when this feature would add to the product version of MKL. 

best, Gennady 

[Hao_W_]

Hey Kenneth and Gennady

Thank you for your response. Right now I know the multithreading is not here for sparse matrix transpose functions of MKL. I will keep an eye on the future version of MKL for this functionality.

- Hao

[Gennady_F_Intel]

FYI: This optimization has been added into 2-stage API for Sparse BLAS (analyze and execute) , MKL 11.3 beta already contains this optimization. The "old" API of this functionality will not be changed.