GitHub - OSU-Nowlab/HPN-Tutorial: Scripts for various clusters to run network-based computing exercises (IB, MPI, GPU-Aware MPI, DPU Offload, etc.)

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 9 Commits
README.TXT		README.TXT
run_dpu-thor.sh		run_dpu-thor.sh
run_fabtest_bw.sh		run_fabtest_bw.sh
run_fabtest_lat.sh		run_fabtest_lat.sh
run_gpu.sh		run_gpu.sh
run_inline_ib.sh		run_inline_ib.sh
run_inline_mpi.sh		run_inline_mpi.sh
run_mtu.sh		run_mtu.sh
run_omb.sh		run_omb.sh
run_omb_ofi.sh		run_omb_ofi.sh
run_perftest_bw.sh		run_perftest_bw.sh
run_perftest_lat.sh		run_perftest_lat.sh
run_rc_ud.sh		run_rc_ud.sh

Repository files navigation

# Basic IB Tutorial #
=====================

1. Measure InfiniBand Latency
$  srun -N2 run_perftest_lat.sh

2. Measure InfiniBand Bandwidth
$  srun -N2 run_perftest_bw.sh

3. Benchmark MPI Performance
$  srun -N2 run_omb.sh



# Advanced IB Tutorial #
========================

1. Impact of changing MTU size
$  srun -N2 run_mtu.sh 256 4096

2. Impact of changing Inline threshold
$  srun -N2 run_inline.sh 0 128

3. MPI Collective performance with RC/UD
$  srun -N4 run_rc_ud.sh