Skip to content

Commit 62e481e

Browse files
committed
wip: tunercheck action
Signed-off-by: Raghu Raja <[email protected]>
1 parent a97f82b commit 62e481e

File tree

1 file changed

+56
-0
lines changed

1 file changed

+56
-0
lines changed

.github/workflows/tunercheck.yaml

Lines changed: 56 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,56 @@
1+
name: Build Checks
2+
on: [push, pull_request]
3+
permissions:
4+
contents: read
5+
pull-requests: read
6+
7+
jobs:
8+
linux:
9+
runs-on: ubuntu-22.04
10+
steps:
11+
- name: Fetch and Install EFA Installer Dependencies
12+
run: |
13+
curl -O https://efa-installer.amazonaws.com/aws-efa-installer-latest.tar.gz
14+
tar -xf aws-efa-installer-*.tar.gz
15+
pushd aws-efa-installer/
16+
sudo ./efa_installer.sh -y --skip-kmod
17+
popd
18+
19+
- name: Build Plugin
20+
run: |
21+
set -x
22+
23+
export CC="cc"
24+
export CXX="c++"
25+
26+
# actions/checkout@v4 would drop the plugin source in $PWD,
27+
# so go ahead and build it.
28+
./autogen.sh
29+
./configure --with-mpi=/opt/amazon/openmpi \
30+
--with-libfabric=/opt/amazon/efa \
31+
--enable-tests=yes \
32+
--enable-platform-aws \
33+
--with-cuda=/usr/local/cuda/ \
34+
--prefix=$PWD/install
35+
36+
- name: Call `make`
37+
run: make V=1
38+
39+
- name: Call `make install`
40+
run: sudo make install V=1
41+
42+
- name: Install Python Dependencies
43+
run: pip install uv
44+
45+
- name: Tuner Check
46+
run: |
47+
run show-tuner-decisions $PWD/install/lib/libnccl-ofi-tuner.so \
48+
--min-ranks-per-node 1 --max-ranks-per-node 8 \
49+
--min-nnodes 2 --max-nnodes 2048
50+
51+
- name: Upload build logs
52+
if: failure()
53+
uses: actions/upload-artifact@4cec3d8aa04e39d1a68397de0c4cd6fb9dce8ec1 # v4.6.1
54+
with:
55+
name: tuner-decisions
56+
path: tuner-decisions.log

0 commit comments

Comments
 (0)