hybrid-llama/turboquant/.github/workflows/build-turboquant-vulkan.yml

name: Build & Release TurboQuant (Linux x64 Vulkan)

on:
  workflow_dispatch:
  push:
    branches:
      - feature/turboquant-kv-cache
    paths:
      - '.github/workflows/build-turboquant-vulkan.yml'
      - '**/CMakeLists.txt'
      - '**/*.h'
      - '**/*.hpp'
      - '**/*.c'
      - '**/*.cpp'
      - '**/*.comp'
      - '**/*.glsl'

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:
  linux-x64-vulkan:
    runs-on: ubuntu-22.04
    permissions:
      contents: write

    steps:
      - name: Clone
        uses: actions/checkout@v6

      - name: Set version tag
        id: version
        run: |
          SHORT_SHA=$(git rev-parse --short HEAD)
          TAG="turboquant-linux-x64-vulkan-${SHORT_SHA}"
          echo "tag=${TAG}" >> "$GITHUB_OUTPUT"
          echo "short_sha=${SHORT_SHA}" >> "$GITHUB_OUTPUT"          

      - name: ccache
        uses: ggml-org/ccache-action@v1.2.21
        with:
          key: turboquant-linux-x64-vulkan
          evict-old-files: 1d

      - name: Dependencies
        run: |
          wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | sudo apt-key add -
          sudo wget -qO /etc/apt/sources.list.d/lunarg-vulkan-jammy.list https://packages.lunarg.com/vulkan/lunarg-vulkan-jammy.list
          sudo apt-get update -y
          sudo apt-get install -y build-essential mesa-vulkan-drivers vulkan-sdk          

      - name: Build
        id: cmake_build
        run: |
          cmake -B build \
            -DCMAKE_BUILD_TYPE=Release \
            -DCMAKE_INSTALL_RPATH='$ORIGIN' \
            -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON \
            -DGGML_BACKEND_DL=ON \
            -DGGML_NATIVE=OFF \
            -DGGML_CPU_ALL_VARIANTS=ON \
            -DGGML_VULKAN=ON \
            -DLLAMA_CURL=OFF \
            -DLLAMA_OPENSSL=OFF \
            -DLLAMA_BUILD_SERVER=ON \
            -DLLAMA_BUILD_TOOLS=ON \
            -DLLAMA_BUILD_TESTS=OFF \
            -DLLAMA_BUILD_EXAMPLES=OFF
          cmake --build build --config Release -j $(nproc)          

      - 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 bundled dependencies
        run: |
          echo "=== Shared objects in build/bin ==="
          find build/bin -name "*.so*" -printf '%f\n' | sort
          echo "---"
          echo "=== llama-server dynamic dependencies ==="
          ldd build/bin/llama-server || true
          echo "---"
          echo "=== Binary sizes ==="
          ls -lh build/bin/llama-server build/bin/llama-cli          

      - 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/bin -name "*.so*" -exec cp -P {} release/build/bin/ \; 2>/dev/null || true
          cp LICENSE release/build/bin/ 2>/dev/null || true
          cd release
          zip -r ../llama-turboquant-linux-x64-vulkan.zip .
          tar -czf ../llama-turboquant-linux-x64-vulkan.tar.gz .
          cd ..
          ls -lh llama-turboquant-linux-x64-vulkan.zip llama-turboquant-linux-x64-vulkan.tar.gz          

      - name: Upload artifact
        uses: actions/upload-artifact@v4
        with:
          name: llama-turboquant-linux-x64-vulkan
          path: release/
          retention-days: 30

      - name: Create GitHub Release
        uses: softprops/action-gh-release@v2
        with:
          tag_name: ${{ steps.version.outputs.tag }}
          target_commitish: ${{ github.sha }}
          name: "TurboQuant Linux x64 Vulkan (${{ steps.version.outputs.short_sha }})"
          body: |
            ## TurboQuant KV Cache — Linux x64 (Vulkan)

            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`
            - Vulkan backend (works on any Vulkan 1.2+ GPU: NVIDIA / AMD / Intel)
            - Portable CPU backend (`GGML_CPU_ALL_VARIANTS`) with bundled shared libraries (`$ORIGIN` rpath)

            ### Requirements
            A working Vulkan runtime/driver on the host (`libvulkan1` + GPU driver, e.g. `mesa-vulkan-drivers` or the NVIDIA driver).

            ### Usage
            ```bash
            # Option 1: zip
            unzip llama-turboquant-linux-x64-vulkan.zip
            # Option 2: tar.gz
            tar -xzf llama-turboquant-linux-x64-vulkan.tar.gz

            ./build/bin/llama-server -m model.gguf -ngl 99 --cache-type-k turbo3 --cache-type-v turbo3
            ```

            ### For Atomic Chat integration
            Replace the binary at:
            ```
            ~/.config/Atomic Chat/data/llamacpp/backends/<version>/linux-x64/build/bin/llama-server
            ```            
          files: |
            llama-turboquant-linux-x64-vulkan.zip
            llama-turboquant-linux-x64-vulkan.tar.gz            
          draft: false
          prerelease: true