219 lines
8.0 KiB
YAML
219 lines
8.0 KiB
YAML
name: Build & Release TurboQuant (macOS ARM64)
|
|
|
|
on:
|
|
workflow_dispatch:
|
|
push:
|
|
branches:
|
|
- feature/turboquant-kv-cache
|
|
paths:
|
|
- '.github/workflows/build-turboquant-macos.yml'
|
|
- '**/CMakeLists.txt'
|
|
- '**/*.h'
|
|
- '**/*.hpp'
|
|
- '**/*.c'
|
|
- '**/*.cpp'
|
|
- '**/*.metal'
|
|
|
|
concurrency:
|
|
group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
|
|
cancel-in-progress: true
|
|
|
|
env:
|
|
LLAMA_LOG_COLORS: 1
|
|
LLAMA_LOG_PREFIX: 1
|
|
LLAMA_LOG_TIMESTAMPS: 1
|
|
|
|
jobs:
|
|
macOS-arm64-metal:
|
|
runs-on: macos-latest
|
|
permissions:
|
|
contents: write
|
|
|
|
steps:
|
|
- name: Clone
|
|
uses: actions/checkout@v6
|
|
|
|
- name: Set version tag
|
|
id: version
|
|
run: |
|
|
SHORT_SHA=$(git rev-parse --short HEAD)
|
|
TAG="turboquant-macos-arm64-${SHORT_SHA}"
|
|
echo "tag=${TAG}" >> "$GITHUB_OUTPUT"
|
|
echo "short_sha=${SHORT_SHA}" >> "$GITHUB_OUTPUT"
|
|
|
|
- name: Import code signing certificate
|
|
env:
|
|
MACOS_CERTIFICATE_P12: ${{ secrets.MACOS_CERTIFICATE_P12 }}
|
|
MACOS_CERTIFICATE_PASSWORD: ${{ secrets.MACOS_CERTIFICATE_PASSWORD }}
|
|
KEYCHAIN_PASSWORD: ${{ secrets.KEYCHAIN_PASSWORD }}
|
|
run: |
|
|
CERTIFICATE_PATH=$RUNNER_TEMP/build_certificate.p12
|
|
KEYCHAIN_PATH=$RUNNER_TEMP/app-signing.keychain-db
|
|
|
|
echo -n "$MACOS_CERTIFICATE_P12" | base64 --decode -o $CERTIFICATE_PATH
|
|
|
|
security create-keychain -p "$KEYCHAIN_PASSWORD" $KEYCHAIN_PATH
|
|
security set-keychain-settings -lut 21600 $KEYCHAIN_PATH
|
|
security unlock-keychain -p "$KEYCHAIN_PASSWORD" $KEYCHAIN_PATH
|
|
|
|
security import $CERTIFICATE_PATH -P "$MACOS_CERTIFICATE_PASSWORD" \
|
|
-A -t cert -f pkcs12 -k $KEYCHAIN_PATH
|
|
security set-key-partition-list -S apple-tool:,apple: \
|
|
-k "$KEYCHAIN_PASSWORD" $KEYCHAIN_PATH
|
|
security list-keychain -d user -s $KEYCHAIN_PATH
|
|
|
|
echo "KEYCHAIN_PATH=$KEYCHAIN_PATH" >> "$GITHUB_ENV"
|
|
|
|
IDENTITY=$(security find-identity -v -p codesigning $KEYCHAIN_PATH | head -1 | grep -o '".*"' | tr -d '"')
|
|
echo "CODESIGN_IDENTITY=$IDENTITY" >> "$GITHUB_ENV"
|
|
echo "Signing identity: $IDENTITY"
|
|
|
|
- name: Build
|
|
id: cmake_build
|
|
run: |
|
|
sysctl -a
|
|
cmake -B build \
|
|
-DCMAKE_BUILD_TYPE=Release \
|
|
-DBUILD_SHARED_LIBS=OFF \
|
|
-DLLAMA_CURL=OFF \
|
|
-DLLAMA_OPENSSL=OFF \
|
|
-DGGML_METAL=ON \
|
|
-DGGML_METAL_USE_BF16=ON \
|
|
-DGGML_METAL_EMBED_LIBRARY=ON \
|
|
-DLLAMA_BUILD_SERVER=ON \
|
|
-DLLAMA_BUILD_TOOLS=ON \
|
|
-DLLAMA_BUILD_TESTS=OFF \
|
|
-DLLAMA_BUILD_EXAMPLES=OFF
|
|
cmake --build build --config Release -j $(sysctl -n hw.ncpu)
|
|
|
|
- name: Verify turbo3 support
|
|
run: |
|
|
./build/bin/llama-server --help 2>&1 | grep -A2 "cache-type-k" || true
|
|
echo "---"
|
|
file ./build/bin/llama-server
|
|
./build/bin/llama-server --version 2>&1 || true
|
|
|
|
- name: Verify static binary
|
|
run: |
|
|
echo "=== All dynamic dependencies (should be system-only) ==="
|
|
otool -L build/bin/llama-server
|
|
echo "---"
|
|
echo "=== Checking for non-system dylibs (should be empty) ==="
|
|
if otool -L build/bin/llama-server | grep -vE '/usr/lib|/System|llama-server' | grep '\.dylib'; then
|
|
echo "ERROR: Found non-system dynamic dependency!"
|
|
exit 1
|
|
fi
|
|
echo "OK: Only system dylibs — fully self-contained binary"
|
|
echo "---"
|
|
echo "=== Binary size ==="
|
|
ls -lh build/bin/llama-server
|
|
|
|
- name: Sign binaries
|
|
run: |
|
|
for bin in build/bin/llama-server build/bin/llama-cli build/bin/llama-bench build/bin/llama-perplexity; do
|
|
if [ -f "$bin" ]; then
|
|
echo "Signing $bin ..."
|
|
codesign --force --options runtime --timestamp \
|
|
--entitlements .github/entitlements.plist \
|
|
--sign "$CODESIGN_IDENTITY" "$bin"
|
|
codesign --verify --verbose "$bin"
|
|
fi
|
|
done
|
|
for lib in $(find build -name "*.dylib" 2>/dev/null); do
|
|
echo "Signing dylib $lib ..."
|
|
codesign --force --options runtime --timestamp \
|
|
--sign "$CODESIGN_IDENTITY" "$lib"
|
|
done
|
|
|
|
- name: Prepare release archive
|
|
run: |
|
|
mkdir -p release/build/bin
|
|
cp build/bin/llama-server release/build/bin/
|
|
cp build/bin/llama-cli release/build/bin/ 2>/dev/null || true
|
|
cp build/bin/llama-bench release/build/bin/ 2>/dev/null || true
|
|
cp build/bin/llama-perplexity release/build/bin/ 2>/dev/null || true
|
|
find build -name "*.dylib" -exec cp {} release/build/bin/ \; 2>/dev/null || true
|
|
find build -name "*.metal" -path "*/bin/*" -exec cp {} release/build/bin/ \; 2>/dev/null || true
|
|
cd release
|
|
zip -r ../llama-turboquant-macos-arm64.zip .
|
|
tar -czf ../llama-turboquant-macos-arm64.tar.gz .
|
|
cd ..
|
|
ls -lh llama-turboquant-macos-arm64.zip llama-turboquant-macos-arm64.tar.gz
|
|
|
|
- name: Notarize release archive
|
|
env:
|
|
APPLE_ID: ${{ secrets.APPLE_ID }}
|
|
APPLE_ID_PASSWORD: ${{ secrets.APPLE_ID_PASSWORD }}
|
|
APPLE_TEAM_ID: ${{ secrets.APPLE_TEAM_ID }}
|
|
run: |
|
|
echo "=== Notarizing zip archive ==="
|
|
xcrun notarytool submit llama-turboquant-macos-arm64.zip \
|
|
--apple-id "$APPLE_ID" \
|
|
--password "$APPLE_ID_PASSWORD" \
|
|
--team-id "$APPLE_TEAM_ID" \
|
|
--wait --timeout 10m
|
|
echo "=== Notarizing individual binaries (for tar.gz users) ==="
|
|
for bin in build/bin/llama-server build/bin/llama-cli build/bin/llama-bench build/bin/llama-perplexity; do
|
|
if [ -f "$bin" ]; then
|
|
name=$(basename "$bin")
|
|
echo "--- Notarizing $name ---"
|
|
zip -j "${name}.zip" "$bin"
|
|
xcrun notarytool submit "${name}.zip" \
|
|
--apple-id "$APPLE_ID" \
|
|
--password "$APPLE_ID_PASSWORD" \
|
|
--team-id "$APPLE_TEAM_ID" \
|
|
--wait --timeout 10m
|
|
rm "${name}.zip"
|
|
fi
|
|
done
|
|
echo "=== All notarization complete ==="
|
|
|
|
- name: Clean up keychain
|
|
if: always()
|
|
run: |
|
|
security delete-keychain $KEYCHAIN_PATH 2>/dev/null || true
|
|
|
|
- name: Upload artifact
|
|
uses: actions/upload-artifact@v4
|
|
with:
|
|
name: llama-turboquant-macos-arm64
|
|
path: release/
|
|
retention-days: 30
|
|
|
|
- name: Create GitHub Release
|
|
uses: softprops/action-gh-release@v2
|
|
with:
|
|
tag_name: ${{ steps.version.outputs.tag }}
|
|
target_commitish: ${{ github.sha }}
|
|
name: "TurboQuant macOS ARM64 (${{ steps.version.outputs.short_sha }})"
|
|
body: |
|
|
## TurboQuant KV Cache — macOS ARM64 (Metal)
|
|
|
|
Built from `feature/turboquant-kv-cache` branch at commit `${{ steps.version.outputs.short_sha }}`.
|
|
|
|
### What's included
|
|
- `llama-server` with `--cache-type-k turbo3` / `turbo4` support
|
|
- `llama-cli`, `llama-bench`, `llama-perplexity`
|
|
- Metal backend with BF16 + embedded shader library
|
|
|
|
### Usage
|
|
```bash
|
|
# Option 1: zip (notarized + stapled)
|
|
unzip llama-turboquant-macos-arm64.zip
|
|
# Option 2: tar.gz
|
|
tar -xzf llama-turboquant-macos-arm64.tar.gz
|
|
|
|
./build/bin/llama-server -m model.gguf --cache-type-k turbo3 --cache-type-v turbo3
|
|
```
|
|
|
|
### For Atomic Chat integration
|
|
Replace the binary at:
|
|
```
|
|
~/Library/Application Support/Atomic Chat/data/llamacpp/backends/<version>/macos-arm64/build/bin/llama-server
|
|
```
|
|
files: |
|
|
llama-turboquant-macos-arm64.zip
|
|
llama-turboquant-macos-arm64.tar.gz
|
|
draft: false
|
|
prerelease: true
|