Hi Freddie, Thank you for the detail information.
I set the environment variable "PYFR_XSMM_LIBRARY_PATH" in .bashrc export PYFR_XSMM_LIBRARY_PATH=$HOME/FR/libxmm/lib In libxmm/lib, there are dynamic libraries like libxsmm*.so. Then, I set "[backend-openmp]" section in .ini file as described below. [backend-openmp] cc = icc cblas = /opt/intel/compilers_and_libraries_2017.2.174/linux/mkl/lib/intel64_lin_mic/libmkl_rt.so I got an error "KeyError: "'mul' has no providers"" when I execute PyFR. The error could be related to wrong cblas library pass. In mkl lib directory, there are many libraries. What dynamic library I should set in cblas section? libmkl_ao_worker.so libmkl_cdft_core.so libmkl_intel_thread.so libmkl_sequential.a libmkl_blacs_intelmpi_ilp64.a libmkl_core.a libmkl_lapack95_ilp64.a libmkl_sequential.so libmkl_blacs_intelmpi_ilp64.so libmkl_core.so libmkl_lapack95_lp64.a libmkl_tbb_thread.a libmkl_blacs_intelmpi_lp64.a libmkl_intel_ilp64.a libmkl_rt.so libmkl_tbb_thread.so libmkl_blacs_intelmpi_lp64.so libmkl_intel_ilp64.so libmkl_scalapack_ilp64.a locale libmkl_blas95_ilp64.a libmkl_intel_lp64.a libmkl_scalapack_ilp64.so libmkl_blas95_lp64.a libmkl_intel_lp64.so libmkl_scalapack_lp64.a libmkl_cdft_core.a libmkl_intel_thread.a libmkl_scalapack_lp64.so Best regards, Tom Hi Tom, > > On 06/06/17 15:47, Tom Vilsack wrote: > > Thank you for your reply. > > > > I have installed libxsmm according to commands as described below. > > > > make STATIC=0 BLAS=0 > > make PREFIX=<libxsmm install path> STATIC=0 BLAS=0 install > > > > The make process finished without error. Then, I set parameter in > > "backend-openmp" section in ini file. > > > > [backend-openmp] > > cc = icc > > cblas = <libxsmm install path>/lib/libxsmm.so > > So libxsmm is not a cblas library (that is MKL). All you need to do is > ensure that libxsmm is somewhere where PyFR can find it. If need be you > can point PyFR to the library by doing > > export PYFR_XSMM_LIBRARY_PATH=/path/here/to/lib.so > > > Are the setting parameters correct? Xeon Phi 7250 have 68 physical > > cores, and each core can run 4 threads. Therefore, I set > > $OMP_NUM_THREADS=272. > > For best performance you want one thread per core. > > > > Could you tell me meanings of "gimmik-max-nnz", "libxsmm-block-sz" and > > "libxsmm-max-sz" in "backend-openmp" section in ini file? > > The first parameter decides the cut-off point at which we will no longer > use the GiMMiK matrix multiplication library. The second parameter > controls the block size for libxsmm, although performance does not > appear to be too sensitive to its value. The final parameter decides > the cut-off point at which we will no longer use libxsmm for matrix > multiplications. The hierarchy is: > > 1. libxsmm (if available and size < libxsmm-max-sz) > 2. GiMMiK (else if nonzeros < gimmik-max-nnz) > 3. cblas (else) > > The defaults in PyFR are not too bad, but you may be able to get 20-30% > out for certain test cases by playing around with these parameters. The > key is to ensure that libxsmm is available, this is especially important > at lower polynomial orders. > > Regards, Freddie. > -- You received this message because you are subscribed to the Google Groups "PyFR Mailing List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send an email to [email protected]. Visit this group at https://groups.google.com/group/pyfrmailinglist. For more options, visit https://groups.google.com/d/optout.
