From 2cbb968101e2bfc7d3a7f0f1826c0189355de6fe Mon Sep 17 00:00:00 2001 From: Sylvain Jeaugey Date: Tue, 3 Jan 2023 08:47:43 +0100 Subject: [PATCH] Update README.md Improve MPI example to avoid confusion of number of processes / total number of GPUs. https://github.com/NVIDIA/nccl-tests/issues/54#issuecomment-1212023369 --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index bff6433b89..12be254542 100644 --- a/README.md +++ b/README.md @@ -29,9 +29,9 @@ Run on 8 GPUs (`-g 8`), scanning from 8 Bytes to 128MBytes : $ ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 8 ``` -Run with MPI on 40 processes (potentially on multiple nodes) with 4 GPUs each : +Run with MPI on 10 processes (potentially on multiple nodes) with 4 GPUs each, for a total of 40 GPUs: ```shell -$ mpirun -np 40 ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 4 +$ mpirun -np 10 ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 4 ``` ### Performance