https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/found-answer-and-a-bug-maybe/m-p/937085#M14184 I think this is an mkl bug:<BR /><BR />I don't set OMP_NUM_THREADS<BR />My code uses the omp_set_num_threads()<BR />It seems though that unless OMP_NUM_THREADS is set to something at the beggining of the program it won't honor any future calls to omp_set_num_threads()<BR />Now, if I take out the call to the MKL function, the plain openmp for loop will actually be parallelized well.<BR /><BR />[jpuig@altix jpuig]$ export -n OMP_NUM_THREADS<BR />[jpuig@altix jpuig]$ ./a.out<BR />Elapsed time: 1.8064 - 1 threads loop ratio:1<BR />Elapsed time: 1.79981 - 2 threads loop ratio:0.996353<BR />Elapsed time: 1.85461 - 3 threads loop ratio:1.02669<BR />Elapsed time: 1.82016 - 4 threads loop ratio:1.00762<BR />[jpuig@altix jpuig]$ export OMP_NUM_THREADS=4<BR />[jpuig@altix jpuig]$ ./a.out<BR />Elapsed time: 1.84285 - 1 threads loop ratio:1<BR />Elapsed time: 0.929641 - 2 threads loop ratio:0.504457<BR />Elapsed time: 0.62401 - 3 threads loop ratio:0.338611<BR />Elapsed time: 0.476085 - 4 threads loop ratio:0.258341<BR />[jpuig@altix jpuig]$ Tue, 04 Apr 2006 04:51:26 GMT joan_puig 2006-04-04T04:51:26Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/found-answer-and-a-bug-maybe/m-p/937085#M14184 I think this is an mkl bug:<BR /><BR />I don't set OMP_NUM_THREADS<BR />My code uses the omp_set_num_threads()<BR />It seems though that unless OMP_NUM_THREADS is set to something at the beggining of the program it won't honor any future calls to omp_set_num_threads()<BR />Now, if I take out the call to the MKL function, the plain openmp for loop will actually be parallelized well.<BR /><BR />[jpuig@altix jpuig]$ export -n OMP_NUM_THREADS<BR />[jpuig@altix jpuig]$ ./a.out<BR />Elapsed time: 1.8064 - 1 threads loop ratio:1<BR />Elapsed time: 1.79981 - 2 threads loop ratio:0.996353<BR />Elapsed time: 1.85461 - 3 threads loop ratio:1.02669<BR />Elapsed time: 1.82016 - 4 threads loop ratio:1.00762<BR />[jpuig@altix jpuig]$ export OMP_NUM_THREADS=4<BR />[jpuig@altix jpuig]$ ./a.out<BR />Elapsed time: 1.84285 - 1 threads loop ratio:1<BR />Elapsed time: 0.929641 - 2 threads loop ratio:0.504457<BR />Elapsed time: 0.62401 - 3 threads loop ratio:0.338611<BR />Elapsed time: 0.476085 - 4 threads loop ratio:0.258341<BR />[jpuig@altix jpuig]$ Tue, 04 Apr 2006 04:51:26 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/found-answer-and-a-bug-maybe/m-p/937085#M14184 joan_puig 2006-04-04T04:51:26Z