https://community.intel.com/t5/Intel-MPI-Library/IntelMPI-Starvation-issue/m-p/934159#M2596 <P>Hi Lars,</P> <P>I'm going to close this thread, as it is a duplicate of <A href="http://software.intel.com/en-us/forums/topic/394242">http://software.intel.com/en-us/forums/topic/394242</A></P> <P>Sincerely,<BR /> James Tullos<BR /> Technical Consulting Engineer<BR /> Intel® Cluster Tools</P> Tue, 09 Jul 2013 20:43:08 GMT James_T_Intel 2013-07-09T20:43:08Z https://community.intel.com/t5/Intel-MPI-Library/IntelMPI-Starvation-issue/m-p/934158#M2595 <P>Hi!<BR />Looks like there is a starvation issue when using "wild" permanent recv requests, i.e. recv requests created with&nbsp; MPI_Recv_init(...,...,..,MPI_ANY_SOURCE, ... It appears to me that this requests are more likely matched with send operations of low ranked peers, resulting in a possible starvation for higher ranked peers, especially in large jobs. At least, this is what I also observed with ITCA.</P> <P>I observed this with intel-mpi/64/4.0.3/008</P> <P>Could some Intel developer check or comment on this?</P> <P>Thanks</P> <P></P> Wed, 26 Jun 2013 11:10:11 GMT https://community.intel.com/t5/Intel-MPI-Library/IntelMPI-Starvation-issue/m-p/934158#M2595 Lars_S_1 2013-06-26T11:10:11Z https://community.intel.com/t5/Intel-MPI-Library/IntelMPI-Starvation-issue/m-p/934159#M2596 <P>Hi Lars,</P> <P>I'm going to close this thread, as it is a duplicate of <A href="http://software.intel.com/en-us/forums/topic/394242">http://software.intel.com/en-us/forums/topic/394242</A></P> <P>Sincerely,<BR /> James Tullos<BR /> Technical Consulting Engineer<BR /> Intel® Cluster Tools</P> Tue, 09 Jul 2013 20:43:08 GMT https://community.intel.com/t5/Intel-MPI-Library/IntelMPI-Starvation-issue/m-p/934159#M2596 James_T_Intel 2013-07-09T20:43:08Z