Hi Siegmar, what if you ? mpiexec --host loki:1,exin:1 -np 3 hello_1_mpi
are loki and exin different ? (os, sockets, core) Cheers, Gilles ----- Original Message ----- > Hi, > > I have installed openmpi-v3.x-201705250239-d5200ea on my "SUSE Linux > Enterprise Server 12.2 (x86_64)" with Sun C 5.14 and gcc-7.1.0. > Depending on the machine that I use to start my processes, I have > a problem with "--host" for versions "v3.x" and "master", while > everything works as expected with earlier versions. > > > loki hello_1 111 mpiexec -np 3 --host loki:2,exin hello_1_mpi > ---------------------------------------------------------------------- ---- > There are not enough slots available in the system to satisfy the 3 slots > that were requested by the application: > hello_1_mpi > > Either request fewer slots for your application, or make more slots available > for use. > ---------------------------------------------------------------------- ---- > > > > Everything is ok if I use the same command on "exin". > > exin fd1026 107 mpiexec -np 3 --host loki:2,exin hello_1_mpi > Process 0 of 3 running on loki > Process 1 of 3 running on loki > Process 2 of 3 running on exin > ... > > > > Everything is also ok if I use openmpi-v2.x-201705260340-58c6b3c on " loki". > > loki hello_1 114 which mpiexec > /usr/local/openmpi-2.1.2_64_cc/bin/mpiexec > loki hello_1 115 mpiexec -np 3 --host loki:2,exin hello_1_mpi > Process 0 of 3 running on loki > Process 1 of 3 running on loki > Process 2 of 3 running on exin > ... > > > "exin" is a virtual machine on QEMU so that it uses a slightly different > processor architecture, e.g., it has no L3 cache but larger L2 caches. > > loki fd1026 117 cat /proc/cpuinfo | grep -e "model name" -e "physical id" -e > "cpu cores" -e "cache size" | sort | uniq > cache size : 15360 KB > cpu cores : 6 > model name : Intel(R) Xeon(R) CPU E5-2620 v3 @ 2.40GHz > physical id : 0 > physical id : 1 > > > loki fd1026 118 ssh exin cat /proc/cpuinfo | grep -e "model name" -e " physical > id" -e "cpu cores" -e "cache size" | sort | uniq > cache size : 4096 KB > cpu cores : 6 > model name : Intel Core Processor (Haswell, no TSX) > physical id : 0 > physical id : 1 > > > Any ideas what's different in the newer versions of Open MPI? Is the new > behavior intended? I would be grateful, if somebody can fix the problem, > if "mpiexec -np 3 --host loki:2,exin hello_1_mpi" should print my messages > in versions "3.x" and "master" as well, if the programs are started on any > machine. Do you need anything else? Thank you very much for any help in > advance. > > > Kind regards > > Siegmar > _______________________________________________ > users mailing list > users@lists.open-mpi.org > https://rfd.newmexicoconsortium.org/mailman/listinfo/users > _______________________________________________ users mailing list users@lists.open-mpi.org https://rfd.newmexicoconsortium.org/mailman/listinfo/users
