https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233417#M30399 <P>I use mkl_sparse_spmm on two 80 *80 matrix, I run the function in a loop. The time cost is at 10e-5 seconds level in most cases, but very often,&nbsp; it&nbsp; can up to be 0.1+ second.</P> <P>Why it suddenly slow? Should I always use mkl_sparse_set_mm_hint before mkl_sparse_spmm ?</P> <DIV style="display: inline;">&nbsp;</DIV> <P><LI-WRAPPER></LI-WRAPPER></P> <P>&nbsp;</P> <P>&nbsp;</P> Wed, 02 Dec 2020 08:52:37 GMT charlesqhappy 2020-12-02T08:52:37Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233417#M30399 <P>I use mkl_sparse_spmm on two 80 *80 matrix, I run the function in a loop. The time cost is at 10e-5 seconds level in most cases, but very often,&nbsp; it&nbsp; can up to be 0.1+ second.</P> <P>Why it suddenly slow? Should I always use mkl_sparse_set_mm_hint before mkl_sparse_spmm ?</P> <DIV style="display: inline;">&nbsp;</DIV> <P><LI-WRAPPER></LI-WRAPPER></P> <P>&nbsp;</P> <P>&nbsp;</P> Wed, 02 Dec 2020 08:52:37 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233417#M30399 charlesqhappy 2020-12-02T08:52:37Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233517#M30401 <P>I don't think if mkl_sparse_set_mm_hint will help. I think it is OS related problems. Is that windows os?</P><BR /> Wed, 02 Dec 2020 13:54:31 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233517#M30401 Gennady_F_Intel 2020-12-02T13:54:31Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233814#M30414 <P>Yeah, I'm using windows. And Unity and C++ dll in Unity. The mkl code is in c++ dll.</P> <P>&nbsp;</P> <P>Very strange, what could happen when the single function slow down? I just do a multiplication.</P> Thu, 03 Dec 2020 07:51:09 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Poor-performance-of-mkl-sparse-spmm/m-p/1233814#M30414 charlesqhappy 2020-12-03T07:51:09Z