diff --git a/scripts/benchmark_sbatch_submit.sh b/scripts/benchmark_sbatch_submit.sh index a75647014ad8b785d5e658cc05de8500571c4c1e..d0a67686cec8c3f677718f988dc466a8b599a424 100644 --- a/scripts/benchmark_sbatch_submit.sh +++ b/scripts/benchmark_sbatch_submit.sh @@ -4,32 +4,36 @@ NUM_NODES=1 dim=2 if [ $NUM_NODES -eq 1 ]; then - for nodes in {1..1}; do - for gpus in {1,8}; do - for batch_size in {64,128}; do - for iteration in {1..1}; do - - echo dim ${dim}, nodes ${nodes}, gpus ${gpus}, batch_size ${batch_size}, iteration ${iteration} - # For single node - sbatch -o sbatch_out/benchmark_dim${dim}_nodes${nodes}_gpus${gpus}_batchsize${batch_size}_iteration${iteration}.out scripts/benchmark_single_node.sbatch ${dim} ${nodes} ${gpus} ${batch_size} ${iteration} + for dim in {2,3}; do + for nodes in {1..1}; do + for gpus in {1,8}; do + for batch_size in {64,128}; do + for iteration in {1..1}; do - sleep 1 # pause to be kind to the scheduler + echo dim ${dim}, nodes ${nodes}, gpus ${gpus}, batch_size ${batch_size}, iteration ${iteration} + # For single node + sbatch -o sbatch_out/benchmark_dim${dim}_nodes${nodes}_gpus${gpus}_batchsize${batch_size}_iteration${iteration}.out scripts/benchmark_single_node.sbatch ${dim} ${nodes} ${gpus} ${batch_size} ${iteration} + sleep 1 # pause to be kind to the scheduler + + done done done done done else - for nodes in {2.. $NUM_NODES}; do - for gpus in {8,8}; do - for batch_size in {64,128}; do - for iteration in {1..1}; do + for dim in {2,3}; do + for nodes in {2.. $NUM_NODES}; do + for gpus in {8,8}; do + for batch_size in {64,128}; do + for iteration in {1..1}; do - # For multi node - sbatch -o sbatch_out/benchmark_dim${dim}_nodes${nodes}_gpus${gpus}_batchsize${batch_size}_iteration${iteration}.out scripts/benchmark_multi_node.sbatch ${dim} ${nodes} ${gpus} ${batch_size} ${iteration} + # For multi node + sbatch -o sbatch_out/benchmark_dim${dim}_nodes${nodes}_gpus${gpus}_batchsize${batch_size}_iteration${iteration}.out scripts/benchmark_multi_node.sbatch ${dim} ${nodes} ${gpus} ${batch_size} ${iteration} - sleep 1 # pause to be kind to the scheduler + sleep 1 # pause to be kind to the scheduler + done done done done