|
| 1 | +name: CI Cortex CPP |
| 2 | + |
| 3 | +on: |
| 4 | + push: |
| 5 | + tags: ["v[0-9]+.[0-9]+.[0-9]+-[0-9]+.[0-9]+.[0-9]+"] |
| 6 | + paths: |
| 7 | + [ |
| 8 | + "cortex-cpp/**", |
| 9 | + ] |
| 10 | + workflow_dispatch: |
| 11 | + |
| 12 | +env: |
| 13 | + LLM_MODEL_URL: https://delta.jan.ai/tinyllama-1.1b-chat-v0.3.Q2_K.gguf |
| 14 | + EMBEDDING_MODEL_URL: https://catalog.jan.ai/dist/models/embeds/nomic-embed-text-v1.5.f16.gguf |
| 15 | + |
| 16 | +jobs: |
| 17 | + create-draft-release: |
| 18 | + runs-on: ubuntu-latest |
| 19 | + if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') |
| 20 | + outputs: |
| 21 | + upload_url: ${{ steps.create_release.outputs.upload_url }} |
| 22 | + version: ${{ steps.get_version.outputs.version }} |
| 23 | + permissions: |
| 24 | + contents: write |
| 25 | + steps: |
| 26 | + - name: Extract tag name without v prefix |
| 27 | + id: get_version |
| 28 | + run: | |
| 29 | + echo "VERSION=${GITHUB_REF#refs/tags/v}" >> $GITHUB_ENV && echo "::set-output name=version::${GITHUB_REF#refs/tags/v}" |
| 30 | + env: |
| 31 | + GITHUB_REF: ${{ github.ref }} |
| 32 | + - name: Create Draft Release |
| 33 | + id: create_release |
| 34 | + uses: softprops/action-gh-release@v2 |
| 35 | + with: |
| 36 | + tag_name: ${{ github.ref_name }} |
| 37 | + token: ${{ secrets.GITHUB_TOKEN }} |
| 38 | + name: "${{ env.VERSION }}" |
| 39 | + draft: false |
| 40 | + generate_release_notes: true |
| 41 | + prerelease: true |
| 42 | + |
| 43 | + build-and-test: |
| 44 | + runs-on: ${{ matrix.runs-on }} |
| 45 | + needs: [create-draft-release] |
| 46 | + timeout-minutes: 40 |
| 47 | + strategy: |
| 48 | + fail-fast: false |
| 49 | + matrix: |
| 50 | + include: |
| 51 | + - os: "linux" |
| 52 | + name: "amd64-avx2" |
| 53 | + runs-on: "ubuntu-18-04" |
| 54 | + cmake-flags: "-DLLAMA_AVX2=ON -DLLAMA_NATIVE=OFF" |
| 55 | + run-e2e: true |
| 56 | + |
| 57 | + - os: "linux" |
| 58 | + name: "amd64-avx" |
| 59 | + runs-on: "ubuntu-18-04" |
| 60 | + cmake-flags: "-DLLAMA_AVX2=OFF -DLLAMA_NATIVE=OFF" |
| 61 | + run-e2e: false |
| 62 | + |
| 63 | + - os: "linux" |
| 64 | + name: "amd64-avx512" |
| 65 | + runs-on: "ubuntu-18-04" |
| 66 | + cmake-flags: "-DLLAMA_AVX512=ON -DLLAMA_NATIVE=OFF" |
| 67 | + run-e2e: false |
| 68 | + |
| 69 | + - os: "linux" |
| 70 | + name: "amd64-vulkan" |
| 71 | + runs-on: "ubuntu-18-04-cuda-11-7" |
| 72 | + cmake-flags: "-DLLAMA_VULKAN=ON -DLLAMA_NATIVE=OFF" |
| 73 | + run-e2e: false |
| 74 | + |
| 75 | + - os: "linux" |
| 76 | + name: "amd64-cuda-11-7" |
| 77 | + runs-on: "ubuntu-18-04-cuda-11-7" |
| 78 | + cmake-flags: "-DCUDA_11_7=ON -DLLAMA_NATIVE=OFF -DLLAMA_CUDA=ON" |
| 79 | + run-e2e: false |
| 80 | + |
| 81 | + - os: "linux" |
| 82 | + name: "amd64-cuda-12-0" |
| 83 | + runs-on: "ubuntu-18-04-cuda-12-0" |
| 84 | + cmake-flags: "-DCUDA_12_0=ON -DLLAMA_NATIVE=OFF -DLLAMA_CUDA=ON" |
| 85 | + run-e2e: false |
| 86 | + |
| 87 | + - os: "mac" |
| 88 | + name: "amd64" |
| 89 | + runs-on: "macos-13" |
| 90 | + cmake-flags: "" |
| 91 | + run-e2e: true |
| 92 | + |
| 93 | + - os: "mac" |
| 94 | + name: "arm64" |
| 95 | + runs-on: "macos-latest" |
| 96 | + cmake-flags: "-DMAC_ARM64=ON" |
| 97 | + run-e2e: false |
| 98 | + |
| 99 | + - os: "windows" |
| 100 | + name: "amd64-avx2" |
| 101 | + runs-on: "windows-cuda-12-0" |
| 102 | + cmake-flags: "-DLLAMA_AVX2=ON -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 103 | + run-e2e: true |
| 104 | + |
| 105 | + - os: "windows" |
| 106 | + name: "amd64-avx" |
| 107 | + runs-on: "windows-cuda-12-0" |
| 108 | + cmake-flags: "-DLLAMA_AVX2=OFF -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 109 | + run-e2e: false |
| 110 | + |
| 111 | + - os: "windows" |
| 112 | + name: "amd64-avx512" |
| 113 | + runs-on: "windows-cuda-12-0" |
| 114 | + cmake-flags: "-DLLAMA_AVX512=ON -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_BLAS=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 115 | + run-e2e: false |
| 116 | + |
| 117 | + - os: "windows" |
| 118 | + name: "amd64-vulkan" |
| 119 | + runs-on: "windows-cuda-12-0" |
| 120 | + cmake-flags: "-DLLAMA_VULKAN=ON -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 121 | + run-e2e: false |
| 122 | + |
| 123 | + - os: "windows" |
| 124 | + name: "amd64-avx2-cuda-12-0" |
| 125 | + runs-on: "windows-cuda-12-0" |
| 126 | + cmake-flags: "-DLLAMA_AVX2=ON -DLLAMA_NATIVE=OFF -DCUDA_12_0=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 127 | + run-e2e: false |
| 128 | + |
| 129 | + - os: "windows" |
| 130 | + name: "amd64-avx-cuda-12-0" |
| 131 | + runs-on: "windows-cuda-12-0" |
| 132 | + cmake-flags: "-DLLAMA_AVX2=OFF -DLLAMA_NATIVE=OFF -DCUDA_12_0=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 133 | + run-e2e: false |
| 134 | + |
| 135 | + - os: "windows" |
| 136 | + name: "amd64-avx512-cuda-12-0" |
| 137 | + runs-on: "windows-cuda-12-0" |
| 138 | + cmake-flags: "-DLLAMA_AVX512=ON -DLLAMA_NATIVE=OFF -DCUDA_12_0=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 139 | + run-e2e: false |
| 140 | + |
| 141 | + - os: "windows" |
| 142 | + name: "amd64-avx2-cuda-11-7" |
| 143 | + runs-on: "windows-cuda-11-7" |
| 144 | + cmake-flags: "-DLLAMA_AVX2=ON -DLLAMA_NATIVE=OFF -DCUDA_11_7=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 145 | + run-e2e: false |
| 146 | + |
| 147 | + - os: "windows" |
| 148 | + name: "amd64-avx-cuda-11-7" |
| 149 | + runs-on: "windows-cuda-11-7" |
| 150 | + cmake-flags: "-DLLAMA_AVX2=OFF -DLLAMA_NATIVE=OFF -DCUDA_11_7=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 151 | + run-e2e: false |
| 152 | + - os: "windows" |
| 153 | + name: "amd64-avx512-cuda-11-7" |
| 154 | + runs-on: "windows-cuda-11-7" |
| 155 | + cmake-flags: "-DLLAMA_AVX512=ON -DLLAMA_NATIVE=OFF -DCUDA_11_7=ON -DLLAMA_BUILD_SERVER=ON -DLLAMA_CUDA=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_BUILD_TYPE=RELEASE" |
| 156 | + run-e2e: false |
| 157 | + |
| 158 | + steps: |
| 159 | + - name: Clone |
| 160 | + id: checkout |
| 161 | + uses: actions/checkout@v3 |
| 162 | + with: |
| 163 | + submodules: recursive |
| 164 | + |
| 165 | + - uses: actions/setup-dotnet@v3 |
| 166 | + if: runner.os == 'Windows' |
| 167 | + with: |
| 168 | + dotnet-version: "8.0.x" |
| 169 | + |
| 170 | + - name: Install choco on Windows |
| 171 | + if: runner.os == 'Windows' |
| 172 | + run: | |
| 173 | + choco install make -y |
| 174 | +
|
| 175 | + - name: Get Cer for code signing |
| 176 | + if: runner.os == 'macOS' |
| 177 | + run: base64 -d <<< "$CODE_SIGN_P12_BASE64" > /tmp/codesign.p12 |
| 178 | + shell: bash |
| 179 | + env: |
| 180 | + CODE_SIGN_P12_BASE64: ${{ secrets.CODE_SIGN_P12_BASE64 }} |
| 181 | + |
| 182 | + - uses: apple-actions/import-codesign-certs@v2 |
| 183 | + if: runner.os == 'macOS' |
| 184 | + with: |
| 185 | + p12-file-base64: ${{ secrets.CODE_SIGN_P12_BASE64 }} |
| 186 | + p12-password: ${{ secrets.CODE_SIGN_P12_PASSWORD }} |
| 187 | + |
| 188 | + - name: Build |
| 189 | + run: | |
| 190 | + cd cortex-cpp |
| 191 | + make build CMAKE_EXTRA_FLAGS="${{ matrix.cmake-flags }}" |
| 192 | +
|
| 193 | + - name: Pre-package |
| 194 | + run: | |
| 195 | + cd cortex-cpp |
| 196 | + make pre-package |
| 197 | +
|
| 198 | + - name: Code Signing macOS |
| 199 | + if: runner.os == 'macOS' |
| 200 | + run: | |
| 201 | + cd cortex-cpp |
| 202 | + make codesign CODE_SIGN=true DEVELOPER_ID="${{ secrets.DEVELOPER_ID }}" |
| 203 | +
|
| 204 | + - name: Code Signing Windows |
| 205 | + if: runner.os == 'Windows' |
| 206 | + shell: cmd |
| 207 | + run: | |
| 208 | + cd cortex-cpp |
| 209 | + set PATH=%PATH%;%USERPROFILE%\.dotnet\tools |
| 210 | + make codesign CODE_SIGN=true AZURE_KEY_VAULT_URI="${{ secrets.AZURE_KEY_VAULT_URI }}" AZURE_CLIENT_ID="${{ secrets.AZURE_CLIENT_ID }}" AZURE_TENANT_ID="${{ secrets.AZURE_TENANT_ID }}" AZURE_CLIENT_SECRET="${{ secrets.AZURE_CLIENT_SECRET }}" AZURE_CERT_NAME="${{ secrets.AZURE_CERT_NAME }}" |
| 211 | + |
| 212 | + |
| 213 | + - name: Package |
| 214 | + run: | |
| 215 | + cd cortex-cpp |
| 216 | + make package |
| 217 | +
|
| 218 | + - name: Run e2e testing |
| 219 | + if: ${{ matrix.run-e2e }} |
| 220 | + run: | |
| 221 | + cd cortex-cpp |
| 222 | + make run-e2e-test RUN_TESTS=true LLM_MODEL_URL=${{ env.LLM_MODEL_URL }} EMBEDDING_MODEL_URL=${{ env.EMBEDDING_MODEL_URL }} |
| 223 | +
|
| 224 | + - name: Upload Artifact |
| 225 | + uses: actions/upload-artifact@v2 |
| 226 | + with: |
| 227 | + name: cortex-cpp-${{ matrix.os }}-${{ matrix.name }} |
| 228 | + path: ./cortex-cpp/cortex-cpp |
| 229 | + |
| 230 | + - uses: actions/upload-release-asset@v1.0.1 |
| 231 | + if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/') |
| 232 | + env: |
| 233 | + GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} |
| 234 | + with: |
| 235 | + upload_url: ${{ needs.create-draft-release.outputs.upload_url }} |
| 236 | + asset_path: ./cortex-cpp/cortex-cpp.tar.gz |
| 237 | + asset_name: cortex-cpp-${{ needs.create-draft-release.outputs.version }}-${{ matrix.os }}-${{ matrix.name }}.tar.gz |
| 238 | + asset_content_type: application/gzip |
0 commit comments