run 8MiB and unaligned bw tests on cooley

- also normalize test order and labels with other platforms

perf-regression/cooley/margo-regression-ofi-rxm.qsub

@@ -18,25 +18,41 @@ export LD_LIBRARY_PATH="/soft/compilers/gcc/7.1.0/lib64:$LD_LIBRARY_PATH"
 # echo "## MPI (one way, double the latency for round trip):"
 # mpirun -f $COBALT_NODEFILE -n 2 ./osu_latency
 
+sleep 1
+
+echo "## Margo OFI/VERBS (round trip):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-latency -i 100000 -n "ofi+verbs;ofi_rxm://"
+echo "## Margo OFI/VERBS (bw, 1MiB):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20
+echo "## Margo OFI/VERBS (bw, 1MiB, 8x concurrency):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20
+echo "## Margo OFI/VERBS (bw, 8MiB):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 8388608 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20
+echo "## Margo OFI/VERBS (bw, 8MiB, 8x concurrency):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 8388608 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20
+echo "## Margo OFI/VERBS (bw, 1MB unaligned):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1000000 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20
+echo "## Margo OFI/VERBS (bw, 1MB unaligned, 8x concurrency):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1000000 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20
+
 sleep 1 
 
 echo "## Margo OFI/VERBS (round trip, Hg busy spin):"
 mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-latency -i 100000 -n "ofi+verbs;ofi_rxm://mlx5_0:3339" -t 0,0
-echo "## Margo OFI/VERBS (bw, Hg busy spin):"
+echo "## Margo OFI/VERBS (bw, 1MiB, Hg busy spin):"
 mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://mlx5_0:3339" -c 1 -D 20 -t 0,0
-echo "## Margo OFI/VERBS (bw, 8x concurrency, Hg busy spin):"
+echo "## Margo OFI/VERBS (bw, 1MiB, 8x concurrency, Hg busy spin):"
 mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://mlx5_0:3339" -c 8 -D 20 -t 0,0
+echo "## Margo OFI/VERBS (bw, 8MiB, Hg busy spin):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 8388608 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20 -t 0,0
+echo "## Margo OFI/VERBS (bw, 8MiB, 8x concurrency, Hg busy spin):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 8388608 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20 -t 0,0
+echo "## Margo OFI/VERBS (bw, 1MB unaligned, Hg busy spin):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1000000 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20 -t 0,0
+echo "## Margo OFI/VERBS (bw, 1MB unaligned, 8x concurrency, Hg busy spin):"
+mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1000000 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20 -t 0,0
 
 
-sleep 1
-
-echo "## Margo OFI/VERBS (round trip):"
-mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-latency -i 100000 -n "ofi+verbs;ofi_rxm://"
-echo "## Margo OFI/VERBS (bw):"
-mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://" -c 1 -D 20
-echo "## Margo OFI/VERBS (bw, 8x concurrency):"
-mpirun -f $COBALT_NODEFILE -n 2 ./margo-p2p-bw -x 1048576 -n "ofi+verbs;ofi_rxm://" -c 8 -D 20
-
 
 # echo "## Mercury-runner CCI/VERBS (round trip):"
 # mpirun -f $COBALT_NODEFILE -n 2 ./mercury-runner -q -c 100000 -l 1 -m c -M -d `pwd` 1 h0=verbs;ofi_rxm:// h1

perf-regression/cooley/run-regression-ofi-rxm.sh

@@ -99,7 +99,7 @@ echo "=== JOB DONE, COLLECTING AND SENDING RESULTS ==="
 # gather output, strip out funny characters, mail
 cat $JOBID.* > combined.$JOBID.txt
 #dos2unix combined.$JOBID.txt
-mailx -s "margo-regression (cooley, ofi rxm)" sds-commits@lists.mcs.anl.gov < combined.$JOBID.txt
+mailx -s "margo-regression (cooley)" sds-commits@lists.mcs.anl.gov < combined.$JOBID.txt
 
 cd /tmp
 rm -rf $SANDBOX