mirror of
https://github.com/microsoft/mscclpp.git
synced 2026-05-11 17:00:22 +00:00
a707273701
Reorganize current native algorithm implementation and DSL algorithm implementation. Provide unified API for DSL algo and native algo and provide interface to tune the algo Provide interface for pytorch integration with native API and DSL --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: chhwang <8018170+chhwang@users.noreply.github.com>
53 lines
1.5 KiB
YAML
53 lines
1.5 KiB
YAML
name: UnitTest
|
|
|
|
on: workflow_dispatch
|
|
|
|
jobs:
|
|
UnitTest:
|
|
runs-on: [ self-hosted, A100 ]
|
|
defaults:
|
|
run:
|
|
shell: bash
|
|
timeout-minutes: 30
|
|
strategy:
|
|
matrix:
|
|
cuda: [ cuda11.8, cuda12.2 ]
|
|
|
|
container:
|
|
image: "ghcr.io/microsoft/mscclpp/mscclpp:base-dev-${{ matrix.cuda }}"
|
|
options: --privileged --ipc=host --gpus=all --ulimit memlock=-1:-1
|
|
|
|
steps:
|
|
- name: Checkout
|
|
uses: actions/checkout@v4
|
|
|
|
- name: Build
|
|
run: |
|
|
mkdir build && cd build
|
|
cmake -DCMAKE_BUILD_TYPE=Release ..
|
|
make -j
|
|
working-directory: ${{ github.workspace }}
|
|
|
|
- name: LockGPUClock
|
|
run: |
|
|
sudo nvidia-smi -pm 1
|
|
for i in $(seq 0 $(( $(nvidia-smi -L | wc -l) - 1 ))); do
|
|
sudo nvidia-smi -ac $(nvidia-smi --query-gpu=clocks.max.memory,clocks.max.sm --format=csv,noheader,nounits -i $i | sed 's/\ //') -i $i
|
|
done
|
|
|
|
- name: UnitTests
|
|
run: |
|
|
./build/bin/unit_tests
|
|
|
|
- name: MpUnitTests
|
|
run: |
|
|
set -e
|
|
mpirun --allow-run-as-root -tag-output -np 2 ./build/bin/mp_unit_tests
|
|
mpirun --allow-run-as-root -tag-output -np 4 ./build/bin/mp_unit_tests
|
|
mpirun --allow-run-as-root -tag-output -np 8 ./build/bin/mp_unit_tests
|
|
|
|
- name: PyTests
|
|
run: |
|
|
set -e
|
|
mpirun --allow-run-as-root -tag-output -np 8 $(which pytest) ./python/test/test_mscclpp.py -x
|