https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MPI-Win-unlock-all-performance/m-p/1162871#M7958 <P>I have been run<SPAN style="font-size: 1em;">ning some performance tests(Intel 17 and 18) and I have noticed that&nbsp;</SPAN><STRONG><EM>MPI_Win_unlock_all</EM>&nbsp;</STRONG>performance is surprisingly excessive with even a small number MPI processes. I was wondering if I could know the reasons and if this is already known.</P> <P>&nbsp;</P> Fri, 24 Nov 2017 11:56:11 GMT luiceur 2017-11-24T11:56:11Z https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MPI-Win-unlock-all-performance/m-p/1162871#M7958 <P>I have been run<SPAN style="font-size: 1em;">ning some performance tests(Intel 17 and 18) and I have noticed that&nbsp;</SPAN><STRONG><EM>MPI_Win_unlock_all</EM>&nbsp;</STRONG>performance is surprisingly excessive with even a small number MPI processes. I was wondering if I could know the reasons and if this is already known.</P> <P>&nbsp;</P> Fri, 24 Nov 2017 11:56:11 GMT https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MPI-Win-unlock-all-performance/m-p/1162871#M7958 luiceur 2017-11-24T11:56:11Z https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MPI-Win-unlock-all-performance/m-p/1162872#M7959 <P>luiceur, try to use MPI_Win_fence(...) instead of MPI_Win_unlock_all, and then compute the execution time.</P> Sat, 04 Apr 2020 16:12:39 GMT https://community.intel.com/t5/Intel-Moderncode-for-Parallel/MPI-Win-unlock-all-performance/m-p/1162872#M7959 ArthurRatz 2020-04-04T16:12:39Z