Skip to content

Commit 46657d2

Browse files
TroyGardenfacebook-github-bot
authored andcommitted
Add support for FakeProcessGroup for EBC (#2228)
Summary: Pull Request resolved: #2228 # context * use FakeProcessGroup to mimic the multi-process tests * can use `_test_compile_fake_pg_fn` as the single-process VB compile test ``` from torchrec.distributed.tests.test_pt2_multiprocess import _test_compile_fake_pg_fn _test_compile_fake_pg_fn( rank=0, world_size=2, ) ``` reference: D59637444 NOTE: right now only tested for EBC, not sure about other sparse modules like PEA or VLE, which shouldn't be too hard to add similar changes. Differential Revision: D51095381
1 parent 5f8a495 commit 46657d2

File tree

1 file changed

+17
-17
lines changed

1 file changed

+17
-17
lines changed

.github/workflows/unittest_ci.yml

+17-17
Original file line numberDiff line numberDiff line change
@@ -5,18 +5,18 @@ name: Unit Test CI
55

66
on:
77
# TODO: re-enable when GPU unit tests are working
8-
# push:
9-
# paths-ignore:
10-
# - "docs/*"
11-
# - "third_party/*"
12-
# - .gitignore
13-
# - "*.md"
14-
# pull_request:
15-
# paths-ignore:
16-
# - "docs/*"
17-
# - "third_party/*"
18-
# - .gitignore
19-
# - "*.md"
8+
push:
9+
paths-ignore:
10+
- "docs/*"
11+
- "third_party/*"
12+
- .gitignore
13+
- "*.md"
14+
pull_request:
15+
paths-ignore:
16+
- "docs/*"
17+
- "third_party/*"
18+
- .gitignore
19+
- "*.md"
2020
workflow_dispatch:
2121

2222
jobs:
@@ -36,12 +36,12 @@ jobs:
3636
- name: Check ldd --version
3737
run: ldd --version
3838
- name: Checkout
39-
uses: actions/checkout@v2
39+
uses: actions/checkout@v3
4040
- name: Update pip
4141
run: |
4242
sudo yum update -y
4343
sudo yum -y install git python3-pip
44-
sudo pip3 install --upgrade pip
44+
# sudo pip3 install --upgrade pip
4545
- name: Setup conda
4646
run: |
4747
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh
@@ -97,7 +97,7 @@ jobs:
9797
runs-on: ${{ matrix.os }}
9898
strategy:
9999
matrix:
100-
os: [linux.4xlarge.nvidia.gpu]
100+
os: [linux.gcp.a100.large]
101101
python-version: [3.8]
102102
cuda-tag: ["cu11"]
103103
needs: build_on_cpu
@@ -135,7 +135,7 @@ jobs:
135135
sudo lshw -C display
136136
# Checkout the repository to the GitHub Actions runner
137137
- name: Checkout
138-
uses: actions/checkout@v2
138+
uses: actions/checkout@v3
139139
- name: Update pip
140140
run: |
141141
sudo yum update -y
@@ -181,7 +181,7 @@ jobs:
181181
- name: Install TorchRec GPU
182182
run: |
183183
rm -r dist || true
184-
conda run -n build_binary python -m pip install dist/*.whl
184+
conda run -n build_binary python -m pip install *.whl
185185
- name: Test torchrec installation
186186
shell: bash
187187
run: |

0 commit comments

Comments
 (0)