ollama/.github/workflows/release.yaml
Daniel Hiltgen e91ae3d47d
Update ROCm (6.3 linux, 6.2 windows) and CUDA v12.8 (#9304)
* Bump cuda and rocm versions

Update ROCm to linux:6.3 win:6.2 and CUDA v12 to 12.8.
Yum has some silent failure modes, so largely switch to dnf.

* Fix windows build script
2025-02-25 13:47:36 -08:00

480 lines
19 KiB
YAML

name: release
on:
push:
tags:
- 'v*'
env:
CGO_CFLAGS: '-O3'
CGO_CXXFLAGS: '-O3'
jobs:
setup-environment:
runs-on: ubuntu-latest
environment: release
outputs:
GOFLAGS: ${{ steps.goflags.outputs.GOFLAGS }}
steps:
- uses: actions/checkout@v4
- name: Set environment
id: goflags
run: |
echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${GITHUB_REF_NAME#v}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_OUTPUT
darwin-build:
runs-on: macos-13
environment: release
needs: setup-environment
strategy:
matrix:
os: [darwin]
arch: [amd64, arm64]
env:
GOFLAGS: ${{ needs.setup-environment.outputs.GOFLAGS }}
steps:
- uses: actions/checkout@v4
- uses: actions/setup-go@v5
with:
go-version-file: go.mod
- run: |
go build -o dist/ .
env:
GOOS: ${{ matrix.os }}
GOARCH: ${{ matrix.arch }}
CGO_ENABLED: 1
CGO_CPPFLAGS: '-mmacosx-version-min=11.3'
- if: matrix.arch == 'amd64'
run: |
cmake --preset CPU -DCMAKE_OSX_DEPLOYMENT_TARGET=11.3 -DCMAKE_SYSTEM_PROCESSOR=x86_64 -DCMAKE_OSX_ARCHITECTURES=x86_64
cmake --build --parallel --preset CPU
cmake --install build --component CPU --strip --parallel 8
- uses: actions/upload-artifact@v4
with:
name: build-${{ matrix.os }}-${{ matrix.arch }}
path: dist/*
darwin-sign:
runs-on: macos-13
environment: release
needs: darwin-build
steps:
- uses: actions/checkout@v4
- run: |
echo $MACOS_SIGNING_KEY | base64 --decode > certificate.p12
security create-keychain -p password build.keychain
security default-keychain -s build.keychain
security unlock-keychain -p password build.keychain
security import certificate.p12 -k build.keychain -P $MACOS_SIGNING_KEY_PASSWORD -T /usr/bin/codesign
security set-key-partition-list -S apple-tool:,apple:,codesign: -s -k password build.keychain
security set-keychain-settings -lut 3600 build.keychain
env:
MACOS_SIGNING_KEY: ${{ secrets.MACOS_SIGNING_KEY }}
MACOS_SIGNING_KEY_PASSWORD: ${{ secrets.MACOS_SIGNING_KEY_PASSWORD }}
- uses: actions/download-artifact@v4
with:
name: build-darwin-amd64
path: dist/darwin-amd64
- uses: actions/download-artifact@v4
with:
name: build-darwin-arm64
path: dist/darwin-arm64
- run: |
export VERSION=${GITHUB_REF_NAME#v}
./scripts/build_darwin.sh sign macapp
env:
APPLE_IDENTITY: ${{ secrets.APPLE_IDENTITY }}
APPLE_PASSWORD: ${{ secrets.APPLE_PASSWORD }}
APPLE_TEAM_ID: ${{ vars.APPLE_TEAM_ID }}
APPLE_ID: ${{ vars.APPLE_ID }}
SDKROOT: /Applications/Xcode_14.1.0.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk
DEVELOPER_DIR: /Applications/Xcode_14.1.0.app/Contents/Developer
- uses: actions/upload-artifact@v4
with:
name: dist-darwin
path: |
dist/Ollama-darwin.zip
dist/ollama-darwin.tgz
windows-depends:
strategy:
matrix:
os: [windows]
arch: [amd64]
preset: ['CPU']
include:
- os: windows
arch: amd64
preset: 'CUDA 11'
install: https://developer.download.nvidia.com/compute/cuda/11.3.1/local_installers/cuda_11.3.1_465.89_win10.exe
cuda-version: '11.3'
- os: windows
arch: amd64
preset: 'CUDA 12'
install: https://developer.download.nvidia.com/compute/cuda/12.8.0/local_installers/cuda_12.8.0_571.96_windows.exe
cuda-version: '12.8'
- os: windows
arch: amd64
preset: 'ROCm 6'
install: https://download.amd.com/developer/eula/rocm-hub/AMD-Software-PRO-Edition-24.Q4-WinSvr2022-For-HIP.exe
rocm-version: '6.2'
runs-on: ${{ matrix.arch == 'arm64' && format('{0}-{1}', matrix.os, matrix.arch) || matrix.os }}
environment: release
env:
GOFLAGS: ${{ needs.setup-environment.outputs.GOFLAGS }}
steps:
- name: Install system dependencies
run: |
choco install -y --no-progress ccache ninja
ccache -o cache_dir=${{ github.workspace }}\.ccache
- if: startsWith(matrix.preset, 'CUDA ') || startsWith(matrix.preset, 'ROCm ')
id: cache-install
uses: actions/cache/restore@v4
with:
path: |
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA
C:\Program Files\AMD\ROCm
key: ${{ matrix.install }}
- if: startsWith(matrix.preset, 'CUDA ')
name: Install CUDA ${{ matrix.cuda-version }}
run: |
$ErrorActionPreference = "Stop"
if ("${{ steps.cache-install.outputs.cache-hit }}" -ne 'true') {
Invoke-WebRequest -Uri "${{ matrix.install }}" -OutFile "install.exe"
$subpackages = @("cudart", "nvcc", "cublas", "cublas_dev") | Foreach-Object {"${_}_${{ matrix.cuda-version }}"}
Start-Process -FilePath .\install.exe -ArgumentList (@("-s") + $subpackages) -NoNewWindow -Wait
}
$cudaPath = (Resolve-Path "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\*").path
echo "$cudaPath\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
- if: startsWith(matrix.preset, 'ROCm')
name: Install ROCm ${{ matrix.rocm-version }}
run: |
$ErrorActionPreference = "Stop"
if ("${{ steps.cache-install.outputs.cache-hit }}" -ne 'true') {
Invoke-WebRequest -Uri "${{ matrix.install }}" -OutFile "install.exe"
Start-Process -FilePath .\install.exe -ArgumentList '-install' -NoNewWindow -Wait
}
$hipPath = (Resolve-Path "C:\Program Files\AMD\ROCm\*").path
echo "$hipPath\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
echo "CC=$hipPath\bin\clang.exe" | Out-File -FilePath $env:GITHUB_ENV -Append
echo "CXX=$hipPath\bin\clang++.exe" | Out-File -FilePath $env:GITHUB_ENV -Append
- if: matrix.preset == 'CPU'
run: |
echo "CC=clang.exe" | Out-File -FilePath $env:GITHUB_ENV -Append
echo "CXX=clang++.exe" | Out-File -FilePath $env:GITHUB_ENV -Append
- if: ${{ !cancelled() && steps.cache-install.outputs.cache-hit != 'true' }}
uses: actions/cache/save@v4
with:
path: |
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA
C:\Program Files\AMD\ROCm
key: ${{ matrix.install }}
- uses: actions/checkout@v4
- uses: actions/cache@v4
with:
path: ${{ github.workspace }}\.ccache
key: ccache-${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.preset }}
- name: Build target "${{ matrix.preset }}"
run: |
Import-Module 'C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\Common7\Tools\Microsoft.VisualStudio.DevShell.dll'
Enter-VsDevShell -VsInstallPath 'C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise' -SkipAutomaticLocation -DevCmdArguments '-arch=x64 -no_logo'
cmake --preset "${{ matrix.preset }}"
cmake --build --parallel --preset "${{ matrix.preset }}"
cmake --install build --component "${{ startsWith(matrix.preset, 'CUDA ') && 'CUDA' || startsWith(matrix.preset, 'ROCm ') && 'HIP' || 'CPU' }}" --strip --parallel 8
env:
CMAKE_GENERATOR: Ninja
- uses: actions/upload-artifact@v4
with:
name: depends-${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.preset }}
path: dist\*
windows-build:
strategy:
matrix:
os: [windows]
arch: [amd64, arm64]
runs-on: ${{ matrix.arch == 'arm64' && format('{0}-{1}', matrix.os, matrix.arch) || matrix.os }}
environment: release
needs: [setup-environment]
env:
GOFLAGS: ${{ needs.setup-environment.outputs.GOFLAGS }}
steps:
- name: Install AMD64 system dependencies
if: matrix.arch == 'amd64'
run: |
$ErrorActionPreference = "Stop"
Start-Process "C:\msys64\usr\bin\pacman.exe" -ArgumentList @("-S", "--noconfirm", "mingw-w64-clang-x86_64-gcc-compat", "mingw-w64-clang-x86_64-clang") -NoNewWindow -Wait
echo "C:\msys64\usr\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
echo "C:\msys64\clang64\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
- name: Install ARM64 system dependencies
if: matrix.arch == 'arm64'
run: |
$ErrorActionPreference = "Stop"
Set-ExecutionPolicy Bypass -Scope Process -Force
[System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072
iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))
echo "C:\ProgramData\chocolatey\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
choco install -y --no-progress git gzip
echo "C:\Program Files\Git\cmd" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
Invoke-WebRequest -Uri "https://github.com/mstorsjo/llvm-mingw/releases/download/20240619/llvm-mingw-20240619-ucrt-aarch64.zip" -OutFile "${{ runner.temp }}\llvm-mingw-ucrt-aarch64.zip"
Expand-Archive -Path ${{ runner.temp }}\llvm-mingw-ucrt-aarch64.zip -DestinationPath "C:\Program Files\"
$installPath=(Resolve-Path -Path "C:\Program Files\llvm-mingw-*-ucrt-aarch64").path
echo $installPath\bin | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
- uses: actions/checkout@v4
- uses: actions/setup-go@v5
with:
go-version-file: go.mod
- run: |
go build -o dist/${{ matrix.os }}-${{ matrix.arch }}/ .
- if: matrix.arch == 'arm64'
run: |
Invoke-WebRequest -Uri "https://aka.ms/vs/17/release/vc_redist.arm64.exe" -OutFile "dist\windows-arm64\vc_redist.arm64.exe"
- run: |
$env:VERSION='${{ github.ref_name }}' -Replace "v(.*)", '$1'
& .\scripts\build_windows.ps1 buildApp
env:
VCToolsRedistDir: stub
- uses: actions/upload-artifact@v4
with:
name: build-${{ matrix.os }}-${{ matrix.arch }}
path: |
dist\${{ matrix.os }}-${{ matrix.arch }}\*.exe
dist\${{ matrix.os }}-${{ matrix.arch }}-app.exe
windows-sign:
runs-on: windows-2022
environment: release
needs: [windows-depends, windows-build]
steps:
- uses: actions/checkout@v4
- uses: google-github-actions/auth@v2
with:
project_id: ollama
credentials_json: ${{ secrets.GOOGLE_SIGNING_CREDENTIALS }}
- run: |
$ErrorActionPreference = "Stop"
Invoke-WebRequest -Uri "https://go.microsoft.com/fwlink/p/?LinkId=323507" -OutFile "${{ runner.temp }}\sdksetup.exe"
Start-Process "${{ runner.temp }}\sdksetup.exe" -ArgumentList @("/q") -NoNewWindow -Wait
Invoke-WebRequest -Uri "https://github.com/GoogleCloudPlatform/kms-integrations/releases/download/cng-v1.0/kmscng-1.0-windows-amd64.zip" -OutFile "${{ runner.temp }}\plugin.zip"
Expand-Archive -Path "${{ runner.temp }}\plugin.zip" -DestinationPath "${{ runner.temp }}\plugin\"
& "${{ runner.temp }}\plugin\*\kmscng.msi" /quiet
echo "${{ vars.OLLAMA_CERT }}" >ollama_inc.crt
- uses: actions/download-artifact@v4
with:
pattern: build-windows-*
path: dist\
merge-multiple: true
- uses: actions/download-artifact@v4
with:
pattern: depends-windows-amd64-*
path: dist\windows-amd64\
merge-multiple: true
- run: |
& .\scripts\build_windows.ps1 gatherDependencies sign buildInstaller distZip
env:
KEY_CONTAINER: ${{ vars.KEY_CONTAINER }}
- uses: actions/upload-artifact@v4
with:
name: dist-windows
path: |
dist\OllamaSetup.exe
dist\ollama-windows-*.zip
linux-build:
strategy:
matrix:
include:
- os: linux
arch: amd64
target: archive
- os: linux
arch: amd64
target: rocm
- os: linux
arch: arm64
target: archive
runs-on: ${{ matrix.arch == 'arm64' && format('{0}-{1}', matrix.os, matrix.arch) || matrix.os }}
environment: release
needs: setup-environment
env:
GOFLAGS: ${{ needs.setup-environment.outputs.GOFLAGS }}
steps:
- uses: actions/checkout@v4
- uses: docker/setup-buildx-action@v3
- uses: docker/build-push-action@v6
with:
context: .
platforms: ${{ matrix.os }}/${{ matrix.arch }}
target: ${{ matrix.target }}
build-args: |
GOFLAGS=${{ env.GOFLAGS }}
CGO_CFLAGS=${{ env.CGO_CFLAGS }}
CGO_CXXFLAGS=${{ env.CGO_CXXFLAGS }}
outputs: type=local,dest=dist/${{ matrix.os }}-${{ matrix.arch }}
cache-from: type=registry,ref=ollama/ollama:latest
cache-to: type=inline
- run: |
for COMPONENT in bin/* lib/ollama/*; do
case "$COMPONENT" in
bin/ollama) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}.tar.in ;;
lib/ollama/*.so) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}.tar.in ;;
lib/ollama/cuda_v11) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}.tar.in ;;
lib/ollama/cuda_v12) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}.tar.in ;;
lib/ollama/cuda_jetpack5) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}-jetpack5.tar.in ;;
lib/ollama/cuda_jetpack6) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}-jetpack6.tar.in ;;
lib/ollama/rocm) echo $COMPONENT >>ollama-${{ matrix.os }}-${{ matrix.arch }}-rocm.tar.in ;;
esac
done
working-directory: dist/${{ matrix.os }}-${{ matrix.arch }}
- run: |
for ARCHIVE in dist/${{ matrix.os }}-${{ matrix.arch }}/*.tar.in; do
tar c -C dist/${{ matrix.os }}-${{ matrix.arch }} -T $ARCHIVE --owner 0 --group 0 | pigz -9vc >$(basename ${ARCHIVE//.*/}.tgz);
done
- uses: actions/upload-artifact@v4
with:
name: dist-${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.target }}
path: |
*.tgz
# Build each Docker variant (OS, arch, and flavor) separately. Using QEMU is unreliable and slower.
docker-build-push:
strategy:
matrix:
include:
- os: linux
arch: arm64
build-args: |
CGO_CFLAGS
CGO_CXXFLAGS
GOFLAGS
- os: linux
arch: amd64
build-args: |
CGO_CFLAGS
CGO_CXXFLAGS
GOFLAGS
- os: linux
arch: amd64
suffix: '-rocm'
build-args: |
CGO_CFLAGS
CGO_CXXFLAGS
GOFLAGS
FLAVOR=rocm
runs-on: ${{ matrix.arch == 'arm64' && format('{0}-{1}', matrix.os, matrix.arch) || matrix.os }}
environment: release
needs: setup-environment
env:
GOFLAGS: ${{ needs.setup-environment.outputs.GOFLAGS }}
steps:
- uses: actions/checkout@v4
- uses: docker/setup-buildx-action@v3
- uses: docker/login-action@v3
with:
username: ${{ vars.DOCKER_USER }}
password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
- id: build-push
uses: docker/build-push-action@v6
with:
context: .
platforms: ${{ matrix.os }}/${{ matrix.arch }}
build-args: ${{ matrix.build-args }}
outputs: type=image,name=ollama/ollama,push-by-digest=true,name-canonical=true,push=true
cache-from: type=registry,ref=ollama/ollama:latest
cache-to: type=inline
- run: |
mkdir -p ${{ matrix.os }}-${{ matrix.arch }}
echo "${{ steps.build-push.outputs.digest }}" >${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.suffix }}.txt
working-directory: ${{ runner.temp }}
- uses: actions/upload-artifact@v4
with:
name: digest-${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.suffix }}
path: |
${{ runner.temp }}/${{ matrix.os }}-${{ matrix.arch }}-${{ matrix.suffix }}.txt
# Merge Docker images for the same flavor into a single multi-arch manifest
docker-merge-push:
strategy:
matrix:
suffix: ['', '-rocm']
runs-on: linux
environment: release
needs: [docker-build-push]
steps:
- uses: docker/login-action@v3
with:
username: ${{ vars.DOCKER_USER }}
password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
- id: metadata
uses: docker/metadata-action@v4
with:
flavor: |
latest=false
suffix=${{ matrix.suffix }}
images: |
ollama/ollama
tags: |
type=ref,enable=true,priority=600,prefix=pr-,event=pr
type=semver,pattern={{version}}
- uses: actions/download-artifact@v4
with:
pattern: digest-*
path: ${{ runner.temp }}
merge-multiple: true
- run: |
docker buildx imagetools create $(echo '${{ steps.metadata.outputs.json }}' | jq -cr '.tags | map("-t", .) | join(" ")') $(cat *-${{ matrix.suffix }}.txt | xargs printf 'ollama/ollama@%s ')
docker buildx imagetools inspect ollama/ollama:${{ steps.metadata.outputs.version }}
working-directory: ${{ runner.temp }}
# Aggregate all the assets and ship a release
release:
needs: [darwin-sign, windows-sign, linux-build]
runs-on: linux
environment: release
permissions:
contents: write
env:
GH_TOKEN: ${{ github.token }}
steps:
- uses: actions/checkout@v4
- uses: actions/download-artifact@v4
with:
name: dist-darwin
path: dist
- uses: actions/download-artifact@v4
with:
name: dist-windows
path: dist
- uses: actions/download-artifact@v4
with:
pattern: dist-linux-*
path: dist
merge-multiple: true
- run: find . -type f -not -name 'sha256sum.txt' | xargs sha256sum | tee sha256sum.txt
working-directory: dist
- name: Create or update Release
run: |
RELEASE_VERSION="$(echo ${GITHUB_REF_NAME} | cut -f1 -d-)"
echo "Looking for existing release for ${RELEASE_VERSION}"
OLD_TAG=$(gh release ls --json name,tagName | jq -r ".[] | select(.name == \"${RELEASE_VERSION}\") | .tagName")
if [ -n "$OLD_TAG" ]; then
echo "Updating release ${RELEASE_VERSION} to point to new tag ${GITHUB_REF_NAME}"
gh release edit ${OLD_TAG} --tag ${GITHUB_REF_NAME}
else
echo "Creating new release ${RELEASE_VERSION} pointing to tag ${GITHUB_REF_NAME}"
gh release create ${GITHUB_REF_NAME} \
--title ${RELEASE_VERSION} \
--draft \
--generate-notes \
--prerelease
fi
echo "Uploading artifacts for tag ${GITHUB_REF_NAME}"
gh release upload ${GITHUB_REF_NAME} dist/* --clobber