https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/IBM-Benchmark-gets-too-slow-from-16-MPI-processes/m-p/1115776#M24639 <P>Hi, folks</P> <P>I am performing Intel MPI's IMB Benchmark to test the connectivity &amp; bandwidth of my cluster server (CentOS 6.4, Mellanox Infiniband)</P> <P>CMD: $ mpiexec.hydra -genv I_MPI_DEBUG 5 -genv I_MPI_FABIRCS dapl -machinefile machines -ppn 1 -n (# of Procs) IMB-RMA</P> <P>When (# of Procs) is from 2~15, it shows the best performance almost up-to hardware's limitation.</P> <P>However when I tested with (# of Procs) set as 16, the performance drops to x3000 slower !!!</P> <P>Even though I change the topology of the MPI processed, the problem remains.</P> <P>Does anybody know the related issue here?....</P> Wed, 03 Feb 2016 03:18:32 GMT seongyun_k_ 2016-02-03T03:18:32Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/IBM-Benchmark-gets-too-slow-from-16-MPI-processes/m-p/1115776#M24639 <P>Hi, folks</P> <P>I am performing Intel MPI's IMB Benchmark to test the connectivity &amp; bandwidth of my cluster server (CentOS 6.4, Mellanox Infiniband)</P> <P>CMD: $ mpiexec.hydra -genv I_MPI_DEBUG 5 -genv I_MPI_FABIRCS dapl -machinefile machines -ppn 1 -n (# of Procs) IMB-RMA</P> <P>When (# of Procs) is from 2~15, it shows the best performance almost up-to hardware's limitation.</P> <P>However when I tested with (# of Procs) set as 16, the performance drops to x3000 slower !!!</P> <P>Even though I change the topology of the MPI processed, the problem remains.</P> <P>Does anybody know the related issue here?....</P> Wed, 03 Feb 2016 03:18:32 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/IBM-Benchmark-gets-too-slow-from-16-MPI-processes/m-p/1115776#M24639 seongyun_k_ 2016-02-03T03:18:32Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/IBM-Benchmark-gets-too-slow-from-16-MPI-processes/m-p/1115777#M24640 <P>Seongyun, I recommend to address this issue to the MPI forum&nbsp;<SPAN style="font-size: 1em; line-height: 1.5;"><A href="https://software.intel.com/en-us/forums/intel-clusters-and-hpc-technology" target="_blank">https://software.intel.com/en-us/forums/intel-clusters-and-hpc-technology</A></SPAN></P> Thu, 04 Feb 2016 05:46:40 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/IBM-Benchmark-gets-too-slow-from-16-MPI-processes/m-p/1115777#M24640 Gennady_F_Intel 2016-02-04T05:46:40Z