Skip to content

Native JNI S3 PyTorch #1858

Native JNI S3 PyTorch

Native JNI S3 PyTorch #1858

name: Native JNI S3 PyTorch
on:
workflow_dispatch:
inputs:
pt_version:
description: 'pytorch version'
required: false
cuda:
description: 'CUDA version'
required: true
default: 'cu124'
schedule:
- cron: '0 5 * * *'
permissions:
id-token: write
contents: read
jobs:
build-pytorch-jni-linux:
if: github.repository == 'deepjavalibrary/djl'
runs-on: ubuntu-latest
container: nvidia/cuda:12.1.1-cudnn8-devel-ubuntu20.04
steps:
- uses: actions/checkout@v3
- name: Set up JDK 17
uses: actions/setup-java@v4
with:
distribution: 'corretto'
java-version: 17
- uses: actions/cache@v3
with:
path: ~/.gradle/caches
key: ${{ runner.os }}-gradle-${{ hashFiles('*/build.gradle.kts', 'engines/**/build.gradle.kts', 'extensions/**/build.gradle.kts') }}
restore-keys: |
${{ runner.os }}-gradle-
- name: Install Environment
run: |
apt-get update
DEBIAN_FRONTEND=noninteractive apt-get install -y locales curl unzip software-properties-common
apt-get install -y python3-pip python3-distutils
pip3 install awscli cmake
ln -s /usr/local/bin/cmake /usr/bin/cmake
- name: Release JNI prep
shell: bash
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
export PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
echo $PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:compileJNI -Ppt_version=$PYTORCH_VERSION
./gradlew -Pjni -Ppt_version=$PYTORCH_VERSION :integration:test "-Dai.djl.default_engine=PyTorch"
./gradlew :engines:pytorch:pytorch-native:cleanJNI
export TORCH_CUDA_ARCH_LIST="8.0 8.6 8.9 9.0"
CUDA_VERSION=${{ github.event.inputs.cuda }}
export CUDA_VERSION=${CUDA_VERSION:-cu124}
./gradlew :engines:pytorch:pytorch-native:compileJNI -Pcuda=$CUDA_VERSION -Ppt_version=$PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:cleanJNI
./gradlew :engines:pytorch:pytorch-native:compileJNI -Pprecxx11 -Pcuda=$CUDA_VERSION -Ppt_version=$PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:cleanJNI
- name: Configure AWS Credentials
uses: aws-actions/configure-aws-credentials@v4
with:
role-to-assume: arn:aws:iam::425969335547:role/djl-ci-publish-role
aws-region: us-east-2
- name: Copy files to S3 with the AWS CLI
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
aws s3 sync engines/pytorch/pytorch-native/jnilib s3://djl-ai/publish/pytorch/${PYTORCH_VERSION}/jnilib
aws cloudfront create-invalidation --distribution-id E371VB8JQ6NRVY --paths "/pytorch/${PYTORCH_VERSION}/jnilib*"
build-pytorch-jni-precxx11:
if: github.repository == 'deepjavalibrary/djl'
runs-on: ubuntu-latest
container:
image: amazonlinux:2
env:
JAVA_HOME: /usr/lib/jvm/java-17-amazon-corretto
timeout-minutes: 30
needs: create-aarch64-runner
steps:
- name: Install compiler environment
run: |
yum -y update
yum -y install patch git cmake3 python3-devel java-17-amazon-corretto-devel jq
yum -y install gcc10 gcc10-c++
ln -sf /usr/bin/gcc10-gcc /usr/bin/gcc
ln -sf /usr/bin/gcc10-cc /usr/bin/cc
ln -sf /usr/bin/gcc10-g++ /usr/bin/g++
ln -sf /usr/bin/gcc10-c++ /usr/bin/c++
ln -sf /usr/bin/gcc10-ar /usr/bin/ar
ln -sf /usr/bin/cmake3 /usr/bin/cmake
pip3 install awscli --upgrade
- uses: taiki-e/checkout-action@v1
- name: Release JNI prep
run: |
export PATH=$PATH:$JAVA_HOME/bin
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
export PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
echo $PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:compileJNI -Pprecxx11 -Ppt_version=$PYTORCH_VERSION
export PYTORCH_PRECXX11=true
./gradlew -Pjni -Ppt_version=$PYTORCH_VERSION :integration:test "-Dai.djl.default_engine=PyTorch"
- name: Configure AWS Credentials
run: |
oidc_token=$(curl -H "Authorization: bearer $ACTIONS_ID_TOKEN_REQUEST_TOKEN" \
"$ACTIONS_ID_TOKEN_REQUEST_URL&audience=sts.amazonaws.com" | jq -r ".value")
echo "::add-mask::$oidc_token"
read -r AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY AWS_SESSION_TOKEN <<<"$(aws sts assume-role-with-web-identity \
--region "us-east-2" \
--role-arn "arn:aws:iam::425969335547:role/djl-ci-publish-role" \
--role-session-name "native-jni-s3-pytorch-jni-precxx11" \
--web-identity-token "$oidc_token" \
--query "[Credentials.AccessKeyId, Credentials.SecretAccessKey, Credentials.SessionToken]" \
--output text)"
echo "::add-mask::$AWS_ACCESS_KEY_ID"
echo "::add-mask::$AWS_SECRET_ACCESS_KEY"
echo "::add-mask::$AWS_SESSION_TOKEN"
echo "AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID" >> "$GITHUB_ENV"
echo "AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY" >> "$GITHUB_ENV"
echo "AWS_SESSION_TOKEN=$AWS_SESSION_TOKEN" >> "$GITHUB_ENV"
- name: Copy files to S3 with the AWS CLI
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
aws s3 sync engines/pytorch/pytorch-native/jnilib s3://djl-ai/publish/pytorch/${PYTORCH_VERSION}/jnilib
aws cloudfront create-invalidation --distribution-id E371VB8JQ6NRVY --paths "/pytorch/${PYTORCH_VERSION}/jnilib*"
build-pytorch-jni-windows:
if: github.repository == 'deepjavalibrary/djl'
runs-on: windows-latest
steps:
- uses: actions/checkout@v4
- name: Set up JDK 17
uses: actions/setup-java@v4
with:
distribution: 'corretto'
java-version: 17
- uses: actions/cache@v4
with:
path: ~/.gradle/caches
key: ${{ runner.os }}-gradle-${{ hashFiles('*/build.gradle.kts', 'engines/**/build.gradle.kts', 'extensions/**/build.gradle.kts') }}
restore-keys: |
${{ runner.os }}-gradle-
- name: Release CPU JNI
shell: cmd
run: |
call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" amd64
gradlew :engines:pytorch:pytorch-native:compileJNI -Ppt_version=${{ github.event.inputs.pt_version }}
- name: Install CUDA 11.7
shell: cmd
run: |
call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" amd64
curl.exe -L https://developer.download.nvidia.com/compute/cuda/11.7.1/local_installers/cuda_11.7.1_516.94_windows.exe -o cuda.exe
curl.exe -L https://developer.download.nvidia.com/compute/redist/cudnn/v8.5.0/local_installers/11.7/cudnn-windows-x86_64-8.5.0.96_cuda11-archive.zip -o cudnn.zip
cuda.exe -s
unzip.exe cudnn.zip
cp.exe -a cudnn*/include cudnn*/lib cudnn*/bin "C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.7/"
rm.exe -Rf cuda.exe cuda.exe cudnn.zip cudnn*
- name: Release cuda11 JNI
shell: cmd
run: |
call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" amd64
set "CUDA_PATH=%ProgramFiles%\NVIDIA GPU Computing Toolkit\CUDA\v11.7"
set "PATH=%CUDA_PATH%\bin;%CUDA_PATH%\libnvvp;%PATH%"
set CUDA_VERSION=${{ github.event.inputs.cuda }}
if "%CUDA_VERSION%" == "" set CUDA_VERSION=cu124
gradlew :engines:pytorch:pytorch-native:cleanJNI :engines:pytorch:pytorch-native:compileJNI -Pcuda=%CUDA_VERSION% -Ppt_version=${{ github.event.inputs.pt_version }}
- name: Configure Deployment AWS Credentials
uses: aws-actions/configure-aws-credentials@v4
with:
role-to-assume: arn:aws:iam::425969335547:role/djl-ci-publish-role
aws-region: us-east-2
- name: Copy files to S3 with the AWS CLI
shell: bash
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
aws s3 sync engines/pytorch/pytorch-native/jnilib s3://djl-ai/publish/pytorch/${PYTORCH_VERSION}/jnilib
aws cloudfront create-invalidation --distribution-id E371VB8JQ6NRVY --paths "/pytorch/${PYTORCH_VERSION}/jnilib*"
build-pytorch-jni-arm64-macos:
if: github.repository == 'deepjavalibrary/djl'
runs-on: macos-latest-xlarge
steps:
- uses: actions/checkout@v4
- name: Set up JDK 17
uses: actions/setup-java@v4
with:
java-version: 17
distribution: corretto
architecture: aarch64
- uses: actions/cache@v4
with:
path: ~/.gradle/caches
key: ${{ runner.os }}-gradle-${{ hashFiles('*/build.gradle.kts', 'engines/**/build.gradle.kts', 'extensions/**/build.gradle.kts') }}
restore-keys: |
${{ runner.os }}-gradle-
- name: Release JNI prep
shell: bash
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
export PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
echo $PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:compileJNI -Ppt_version=$PYTORCH_VERSION
./gradlew -Pjni -Ppt_version=$PYTORCH_VERSION :integration:test "-Dai.djl.default_engine=PyTorch"
- name: Configure Deployment AWS Credentials
uses: aws-actions/configure-aws-credentials@v4
with:
role-to-assume: arn:aws:iam::425969335547:role/djl-ci-publish-role
aws-region: us-east-2
- name: Copy files to S3 with the AWS CLI
shell: bash
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
aws s3 sync engines/pytorch/pytorch-native/jnilib s3://djl-ai/publish/pytorch/${PYTORCH_VERSION}/jnilib
aws cloudfront create-invalidation --distribution-id E371VB8JQ6NRVY --paths "/pytorch/${PYTORCH_VERSION}/jnilib*"
create-aarch64-runner:
if: github.repository == 'deepjavalibrary/djl'
runs-on: [ self-hosted, scheduler ]
steps:
- name: Create new Graviton instance
id: create_aarch64
run: |
cd /home/ubuntu/djl_benchmark_script/scripts
token=$( curl -X POST -H "Authorization: token ${{ secrets.ACTION_RUNNER_PERSONAL_TOKEN }}" \
https://api.github.com/repos/deepjavalibrary/djl/actions/runners/registration-token \
--fail \
| jq '.token' | tr -d '"' )
./start_instance.sh action_graviton $token djl
outputs:
aarch64_instance_id: ${{ steps.create_aarch64.outputs.action_graviton_instance_id }}
build-pytorch-jni-aarch64:
if: github.repository == 'deepjavalibrary/djl'
runs-on: [ self-hosted, aarch64 ]
container:
image: amazonlinux:2
env:
JAVA_HOME: /usr/lib/jvm/java-17-amazon-corretto.aarch64
timeout-minutes: 30
needs: create-aarch64-runner
steps:
- name: Install compiler environment
run: |
yum -y update
yum -y install patch git cmake3 python3-devel java-17-amazon-corretto-devel jq
yum -y install gcc10 gcc10-c++
ln -sf /usr/bin/gcc10-gcc /usr/bin/gcc
ln -sf /usr/bin/gcc10-cc /usr/bin/cc
ln -sf /usr/bin/gcc10-g++ /usr/bin/g++
ln -sf /usr/bin/gcc10-c++ /usr/bin/c++
ln -sf /usr/bin/gcc10-ar /usr/bin/ar
ln -sf /usr/bin/cmake3 /usr/bin/cmake
pip3 install awscli --upgrade
- uses: taiki-e/checkout-action@v1
- name: Release JNI prep
run: |
export PATH=$PATH:$JAVA_HOME/bin
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
export PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
echo $PYTORCH_VERSION
./gradlew :engines:pytorch:pytorch-native:compileJNI -Pprecxx11 -Ppt_version=$PYTORCH_VERSION
./gradlew -Pjni -Ppt_version=$PYTORCH_VERSION :integration:test "-Dai.djl.default_engine=PyTorch"
- name: Configure AWS Credentials
run: |
oidc_token=$(curl -H "Authorization: bearer $ACTIONS_ID_TOKEN_REQUEST_TOKEN" \
"$ACTIONS_ID_TOKEN_REQUEST_URL&audience=sts.amazonaws.com" | jq -r ".value")
echo "::add-mask::$oidc_token"
read -r AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY AWS_SESSION_TOKEN <<<"$(aws sts assume-role-with-web-identity \
--region "us-east-2" \
--role-arn "arn:aws:iam::425969335547:role/djl-ci-publish-role" \
--role-session-name "native-jni-s3-pytorch-jni-aarch64" \
--web-identity-token "$oidc_token" \
--query "[Credentials.AccessKeyId, Credentials.SecretAccessKey, Credentials.SessionToken]" \
--output text)"
echo "::add-mask::$AWS_ACCESS_KEY_ID"
echo "::add-mask::$AWS_SECRET_ACCESS_KEY"
echo "::add-mask::$AWS_SESSION_TOKEN"
echo "AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID" >> "$GITHUB_ENV"
echo "AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY" >> "$GITHUB_ENV"
echo "AWS_SESSION_TOKEN=$AWS_SESSION_TOKEN" >> "$GITHUB_ENV"
- name: Copy files to S3 with the AWS CLI
run: |
PYTORCH_VERSION=${{ github.event.inputs.pt_version }}
PYTORCH_VERSION=${PYTORCH_VERSION:-$(awk -F '=' '/pytorch/ {gsub(/ ?"/, "", $2); print $2}' gradle/libs.versions.toml)}
aws s3 sync engines/pytorch/pytorch-native/jnilib s3://djl-ai/publish/pytorch/${PYTORCH_VERSION}/jnilib
aws cloudfront create-invalidation --distribution-id E371VB8JQ6NRVY --paths "/pytorch/${PYTORCH_VERSION}/jnilib*"
stop-runners:
if: ${{ github.repository == 'deepjavalibrary/djl' && always() }}
runs-on: [ self-hosted, scheduler ]
needs: [ create-aarch64-runner, build-pytorch-jni-aarch64 ]
steps:
- name: Stop all instances
run: |
cd /home/ubuntu/djl_benchmark_script/scripts
instance_id=${{ needs.create-aarch64-runner.outputs.aarch64_instance_id }}
./stop_instance.sh $instance_id