pmodels
osu_gather/osu_scatter show large variance at 32 nodes 96 ppn
On both current and next-eval images, osu_gather exhibits large variance in latency such as the following:
mpiexec --np 3072 --ppn 96 --cpu-bind verbose,list:1:2:3:4:5:6:7:8:9:10:11:12:13:14:15:16:18:19:20:21:22:23:24:25:26:27:28:29:30:31:32:33:35:36:37:38:39:40:41:42:43:44:45:46:47:48:49:50:53:54:55:56:57:58:59:60:61:62:63:64:65:66:67:68:70:71:72:73:74:75:76:77:78:79:80:81:82:83:84:85:87:88:89:90:91:92:93:94:95:96:97:98:99:100:101:102 --gpu-bind verbose,list:0.0:0.0:0.0:0.0:0.0:0.0:0.0:0.0:0.1:0.1:0.1:0.1:0.1:0.1:0.1:0.1:1.0:1.0:1.0:1.0:1.0:1.0:1.0:1.0:1.1:1.1:1.1:1.1:1.1:1.1:1.1:1.1:2.0:2.0:2.0:2.0:2.0:2.0:2.0:2.0:2.1:2.1:2.1:2.1:2.1:2.1:2.1:2.1:3.0:3.0:3.0:3.0:3.0:3.0:3.0:3.0:3.1:3.1:3.1:3.1:3.1:3.1:3.1:3.1:4.0:4.0:4.0:4.0:4.0:4.0:4.0:4.0:4.1:4.1:4.1:4.1:4.1:4.1:4.1:4.1:5.0:5.0:5.0:5.0:5.0:5.0:5.0:5.0:5.1:5.1:5.1:5.1:5.1:5.1:5.1:5.1 /lus/flare/projects/Aurora_testing/mpi/osu_rfm/run_collective/32/gather-gather_persistent-gatherv-gatherv_persistent/stage/2025-09-12_18-37-26/aurora/compute/PrgEnv-intel/BuildMPIcollective_93bceebc/binaries/osu_gatherv -m 8:8 -i 1000 -x 100 -f -z -d sycl
# OSU MPI-SYCL Gatherv Latency Test v7.5 # Datatype: MPI_CHAR. # Size Avg Latency(us) Min Latency(us) Max Latency(us) Iterations P50 Tail Lat(us) P90 Tail Lat(us) P99 Tail Lat(us) 8 119.12 1.61 124007.70 1000 117.50 123.89 193.58
module load mpich-config/collective-tuning/1024 ; MPIR_CVAR_CH4_PROGRESS_THROTTLE=1 ; mpiexec --np 3072 --ppn 96 --cpu-bind verbose,list:1:2:3:4:5:6:7:8:9:10:11:12:13:14:15:16:18:19:20:21:22:23:24:25:26:27:28:29:30:31:32:33:35:36:37:38:39:40:41:42:43:44:45:46:47:48:49:50:53:54:55:56:57:58:59:60:61:62:63:64:65:66:67:68:70:71:72:73:74:75:76:77:78:79:80:81:82:83:84:85:87:88:89:90:91:92:93:94:95:96:97:98:99:100:101:102 /lus/flare/projects/Aurora_testing/mpi/osu_rfm/run_collective/32/gather-gather_persistent-gatherv-gatherv_persistent/stage/2025-09-14_21-51-44/aurora/compute/PrgEnv-intel/BuildMPIcollective_93bceebc/binaries/osu_gather -m 4096:4096 -i 1000 -x 100 -f -z ; unset MPIR_CVAR_CH4_PROGRESS_THROTTLE ; module unload mpich-config/collective-tuning/1024
# OSU MPI Gather Latency Test v7.5 # Datatype: MPI_CHAR. # Size Avg Latency(us) Min Latency(us) Max Latency(us) Iterations P50 Tail Lat(us) P90 Tail Lat(us) P99 Tail Lat(us) 4096 12.31 0.76 1838.37 1000 12.25 12.36 12.67