MKL Sparse BLAS executor slower than deprecated routines

[SIMKIN__John]

I have several sparse iterative solvers that use the deprecated MKL sparse Blas procedures. After converting the solvers to use the latest sparse BLAS executor functions they all run slower than before. Has anyone else noticed this difference?

                                    Typical times (seconds):
Equations     threads Old_MKL  Executor_MKL
2e5               1           13               17
2e5               6           6                 10
3e6               1           170             250

The solvers used mkl_dcsrtrsv and mkl_dcsrsymv, they were updated to use mkl_sparse_d_trsv and mkl_sparse_d_mv. Double precision and double complex show similar time comparisons.

The comparisons were performed on a windows 10 computer with a Xeon E5 1650 processor, software built using Parallel XE 2019 with intel fortran and C++ compilers.

[Kirill_V_Intel]

Hello,

We haven't heard of this issue. Could you share your workload (matrix) with us so that we can check? Also, which threading did you use (I assume, OpenMP), was it Intel OpenMP or GNU?

Also, did you call mkl_sparse_optimize with appropriate hints before the actual calls to the computational routines?

In other words, if you can give us a simple example (just showing the way you call old SpBLAS and IE SpBLAS), it would speed up our investigation.

Thanks,
Kirill

[Gennady_F_Intel]

John

you may also submit this issue via Intel Online Service Center where you could share the reproducer and all needed details in the case if all of your data are private. 

[SIMKIN__John]

I have now created a small program to demonstrate the problem and submitted it as an issue to the support centre as recommended by F. Gennady.

I will report progress.