checkpoint #95
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
name: Codebase tests (GPU) | |
on: | |
pull_request: | |
push: | |
branches: | |
- eighty-devx-gpu-ci | |
# Allows you to run this workflow manually from the Actions tab | |
workflow_dispatch: | |
env: | |
XLA_FLAGS: --xla_gpu_enable_command_buffer= | |
XLA_PYTHON_CLIENT_PREALLOCATE: "false" | |
XLA_PYTHON_CLIENT_ALLOCATOR: "platform" | |
jobs: | |
test: | |
strategy: | |
matrix: | |
runner: [ParallelHoss] | |
runs-on: ${{ matrix.runner }} | |
steps: | |
- uses: actions/checkout@v4 | |
- name: apt install | |
uses: awalsh128/cache-apt-pkgs-action@latest | |
with: | |
packages: curl | |
version: 1.0 | |
- name: echo | |
run: | | |
echo "ssh..." | |
- name: System debug | |
run: | | |
echo ">nvidia-smi" | |
nvidia-smi | |
echo ">nvcc --version" | |
if ! nvcc --version; then | |
echo "nvcc not found" | |
echo ">find nvcc" | |
sudo find / -name "nvcc" | grep bin | |
fi | |
- name: ssh | |
uses: ryanchapman/gha-ssh@v1 | |
# after the container starts tmate in the background, it will | |
# sleep for 24 hours, so it's important that you set a timeout here | |
# so you don't run up your GitHub Actions bill | |
timeout-minutes: 10 | |
with: | |
# authorized_github_users: required | |
# List of GitHub users who are allowed to ssh into container. | |
# On gha-ssh container start, it downloads the ssh public key(s) for each | |
# user from GitHub and places it in ~/authorized_keys | |
# tmate is started with `-a ~/authorized_keys` to only allow access | |
# to users with possession of the corresponding private ssh keys. | |
authorized_github_users: 'eightysteele' | |
# debug: optional | |
# defaults to `false` if not set here | |
# if debug is set, then tmate is started with `-vvv -F` | |
debug: true | |
# - name: Setup asset caching | |
# id: cache-assets | |
# uses: actions/cache@v4 | |
# with: | |
# path: assets | |
# key: ${{ runner.os }}-${{ matrix.runner }}-assets | |
# restore-keys: ${{ runner.os }}-${{ matrix.runner }}-assets | |
# - name: Setup pixi caching | |
# id: cache-pixi | |
# uses: actions/cache@v4 | |
# with: | |
# path: .pixi | |
# key: ${{ runner.os }}-${{ matrix.runner }}-pixi | |
# restore-keys: ${{ runner.os }}-${{ matrix.runner }}-pixi | |
# - name: System debug | |
# run: | | |
# echo ">nvidia-smi" | |
# nvidia-smi | |
# echo ">nvcc --version" | |
# if ! nvcc --version; then | |
# echo "nvcc not found" | |
# echo ">find nvcc" | |
# sudo find / -name "nvcc" | grep bin | |
# fi | |
# echo ">glxinfo -B" | |
# if ! glxinfo -B; then | |
# echo "glxinfo failed" | |
# fi | |
# echo ">find libGLdispatch.so" | |
# sudo find / -name "*libGLdispatch.so*" | |
# echo ">find nvidia.json" | |
# sudo find / -name "*nvidia*.json" | |
#ldd /lib/x86_64-linux-gnu/libEGL.so.1 | |
#strings /lib/x86_64-linux-gnu/libEGL.so.1 | grep -i 'nvidia\|mesa' | |
# ldconfig -p | grep EGL | |
# ldconfig -p | grep GL | |
# ls -la /usr/share/glvnd/ | |
# ls -la /usr/share/glvnd/egl_vendor.d/ | |
# cat /usr/share/glvnd/egl_vendor.d/50_mesa.json | |
# /usr/share/glvnd/egl_vendor.d/10_nvidia.json | |
# in pixi environment | |
# x86_64-conda-linux-gnu/sysroot/usr/lib64/libGLdispatch.so | |
# sudo sh -c 'echo "{" >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo sh -c 'echo " \"file_format_version\" : \"1.0.0\"," >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo sh -c 'echo " \"ICD\" : {" >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo sh -c 'echo " \"library_path\" : \"libEGL_nvidia.so.0\"" >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo sh -c 'echo " }" >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo sh -c 'echo "}" >> /usr/share/glvnd/egl_vendor.d/10_nvidia.json' | |
# sudo chmod 755 /usr/share/glvnd/egl_vendor.d/10_nvidia.json | |
# cat /usr/share/glvnd/egl_vendor.d/10_nvidia.json | |
# sudo find / -name "*nvidia.so*" | |
# sudo find / -name "*nvidia*" | |
#glxinfo -B | |
# - name: Authenticate gcloud | |
# id: auth | |
# uses: google-github-actions/auth@v2 | |
# with: | |
# credentials_json: "${{ secrets.ARTIFACT_REGISTRY_KEY }}" | |
# - name: Setup gcloud | |
# uses: google-github-actions/setup-gcloud@v2 | |
# - name: Install system environment and project dependencies | |
# env: | |
# GITHUB_CI: "true" | |
# __EGL_VENDOR_LIBRARY_DIRS: "/usr/share/glvnd" | |
# #LD_LIBRARY_PATH: "/usr/lib/x86_64-linux-gnu:$LD_LIBRARY_PATH" | |
# run: | | |
# ./install.sh | |
# source ~/.bashrc | |
# export PATH="/home/runner/.pixi/bin:/home/runner/.local/bin:$PATH" | |
# export CPLUS_INCLUDE_PATH="$GITHUB_WORKSPACE/.pixi/envs/default/targets/x86_64-linux/include" | |
# export LD_LIBRARY_PATH="$GITHUB_WORKSPACE/x86_64-conda-linux-gnu/sysroot/usr/lib64:$GITHUB_WORKSPACE/lib" | |
# #pixi tree | |
# #export LD_LIBRARY_PATH=/lib/x86_64-linux-gnu | |
# env | |
# cp "$GITHUB_WORKSPACE/assets/system/libEGL_nvidia.so.0" "$GITHUB_WORKSPACE/.pixi/envs/default/x86_64-conda-linux-gnu/sysroot/usr/lib64/" | |
# cp "$GITHUB_WORKSPACE/assets/system/libnvidia-glsi.so.555.58.02" "$GITHUB_WORKSPACE/.pixi/envs/default/x86_64-conda-linux-gnu/sysroot/usr/lib64/" | |
# pixi run test-ci | |
# #grep "GL.*\.so" strace_output.txt | |
# #/usr/share/glvnd/egl_vendor.d/50_mesa.json | |
# - name: Run tests | |
# env: | |
# PATH: "/home/runner/.pixi/bin:/home/runner/.local/bin:$PATH" | |
# CONDA_PREFIX: "/home/runner/.pixi/envs/default" | |
# run: pixi run all-tests-ci | |
# - name: Run tests | |
# uses: prefix-dev/setup-pixi@v0.8.1 | |
# env: | |
# PATH: "/home/runner/.pixi/bin:/home/runner/.local/bin:$PATH" | |
# RUNNER_DEBUG: true | |
# with: | |
# pixi-version: v0.25.0 | |
# activate-environment: true | |
# cache: true | |
# - run: | | |
# ls -la .pixi/envs/default | |
#cp assets/system/libEGL.so .pixi/envs/default/x86_64-conda-linux-gnu/sysroot/usr/lib64/ | |
#pixi run test |