https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Performance-issue-when-running-multiple-instances-of-BLAS/m-p/792020#M2292 First, you should know that linking a very simple zgemv program with Intel MKL will automaticallyparallelize the application and use all available cores on the system. Launching more instances will oversubscribe the resources of your system. In other words running two instances of your simply program will mean you have two different programs each trying to use all 12 cores cores. <BR /><BR />Perhaps you want to link the sequential version to run zgemv on just one thread then launch multiple instances. You could also use mkl_set_num_threads() or other available environment variables.<BR /><BR />Perhaps you're aware of all this and you were running into something else, but in that case we could use a little more information on how you're linking your program and how you expect to handle parallelism.<BR /><BR />Todd Fri, 08 Jul 2011 17:36:36 GMT Todd_R_Intel 2011-07-08T17:36:36Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Performance-issue-when-running-multiple-instances-of-BLAS/m-p/792019#M2291 <P>There is significant performance slow-down when I run a simple zgemv program with multiple instances on a machine.<BR />If running a.exe take x time, when I run 4 a.exe at the same time on this machine, each takes 2x+ time. This is on Linux, multi-core (12), large memory machine. The matrix size is about 1400x1400.<BR /><BR />What I can do to improve the performance with multiple instance run here?<BR /><BR />Thanks.</P> Fri, 08 Jul 2011 17:28:53 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Performance-issue-when-running-multiple-instances-of-BLAS/m-p/792019#M2291 missing__zlw 2011-07-08T17:28:53Z https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Performance-issue-when-running-multiple-instances-of-BLAS/m-p/792020#M2292 First, you should know that linking a very simple zgemv program with Intel MKL will automaticallyparallelize the application and use all available cores on the system. Launching more instances will oversubscribe the resources of your system. In other words running two instances of your simply program will mean you have two different programs each trying to use all 12 cores cores. <BR /><BR />Perhaps you want to link the sequential version to run zgemv on just one thread then launch multiple instances. You could also use mkl_set_num_threads() or other available environment variables.<BR /><BR />Perhaps you're aware of all this and you were running into something else, but in that case we could use a little more information on how you're linking your program and how you expect to handle parallelism.<BR /><BR />Todd Fri, 08 Jul 2011 17:36:36 GMT https://community.intel.com/t5/Intel-oneAPI-Math-Kernel-Library/Performance-issue-when-running-multiple-instances-of-BLAS/m-p/792020#M2292 Todd_R_Intel 2011-07-08T17:36:36Z