Skip to content

CI: Update XLA from 1.9 to 1.12 #14013

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 9 commits into from
Aug 5, 2022
Merged

CI: Update XLA from 1.9 to 1.12 #14013

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
1d1cd91
Add 1.12 xla
akihironitta Aug 3, 2022
7d74251
Use 1.11 in tpu tests
akihironitta Aug 3, 2022
bf61a6e
Pin mkl
akihironitta Aug 4, 2022
8ff9376
Update default xla version in dockerfile
akihironitta Aug 4, 2022
89e7ed7
revertme: push
akihironitta Aug 4, 2022
5a61f1f
Revert "revertme: push"
akihironitta Aug 4, 2022
0f3875a
Drop 1.11 xla
akihironitta Aug 5, 2022
67362ce
Update dockers/base-xla/Dockerfile
akihironitta Aug 5, 2022
20c652d
Update doc
akihironitta Aug 5, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion .circleci/config.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,7 +113,7 @@ jobs:
docker:
- image: circleci/python:3.7
environment:
- XLA_VER: 1.9
- XLA_VER: 1.12
- PYTHON_VER: 3.7
- MAX_CHECKS: 1000
- CHECK_SPEEP: 5
Expand Down
2 changes: 1 addition & 1 deletion .github/workflows/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@
| pytorch-lightning (HPUs) | .azure-pipelines/hpu-tests.yml | Run only HPU-specific tests. | HPU | (3.8, 1.10) | linux |
| pytorch-lightning (GPUs) | .azure-pipelines/gpu-tests.yml | Run all CPU and GPU-specific tests, standalone, and examples. Each standalone test needs to be run in separate processes to avoid unwanted interactions between test cases. | GPU | (3.9, 1.12) | linux |
| PyTorchLightning.Benchmark | .azure-pipelines/gpu-benchmark.yml | Run speed/memory benchmarks for parity with pure PyTorch. | GPU | (3.9, 1.12) | linux |
| test-on-tpus | .circleci/config.yml | Run only TPU-specific tests. | TPU | (3.7, 1.9) | linux |
| test-on-tpus | .circleci/config.yml | Run only TPU-specific tests. | TPU | (3.7, 1.12) | linux |

- \*Accelerators used in CI
- GPU: 2 x NVIDIA Tesla V100
Expand Down
2 changes: 1 addition & 1 deletion .github/workflows/cicd-pytorch_dockers.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@ jobs:
matrix:
# the config used in '.circleci/config.yml`'
python_version: ["3.7"]
xla_version: ["1.11"]
xla_version: ["1.12"]
steps:
- uses: actions/checkout@v2
- uses: docker/setup-buildx-action@v2
Expand Down
6 changes: 5 additions & 1 deletion dockers/base-xla/Dockerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,7 @@ LABEL maintainer="Lightning-AI <https://github.com/Lightning-AI>"
# CALL: docker image build -t pytorch-lightning:XLA-image -f dockers/base-xla/Dockerfile . --build-arg PYTHON_VERSION=3.8
ARG PYTHON_VERSION=3.9
ARG CONDA_VERSION=4.9.2
ARG XLA_VERSION=1.11
ARG XLA_VERSION=1.12

SHELL ["/bin/bash", "-c"]
# for skipping configurations
Expand Down Expand Up @@ -92,6 +92,10 @@ RUN \
python --version && \
cd pytorch-lightning && \
pip install -q fire && \
# Pin mkl version to avoid OSError on torch import
# OSError: libmkl_intel_lp64.so.1: cannot open shared object file: No such file or directory
# https://github.com/pytorch/xla/issues/1666
pip install mkl==2021.4.0 && \
Copy link
Contributor

@carmocca carmocca Sep 29, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@akihironitta Why did you pin to this specific release?

I am still seeing the OSError in #14926. I'll try unpinning

Copy link
Contributor Author

@akihironitta akihironitta Sep 29, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If I remember correctly, that version was the latest release that worked in our CI at the time. I believe there's no other reason, so we can try unpinning it.

# drop packages installed with XLA
python .actions/assistant.py requirements_prune_pkgs torch,torchvision && \
# drop unnecessary packages
Expand Down