Add Windows CUDA llama.cpp release asset

Adds a Windows x64 CUDA release asset for the llama.cpp SenseVoice runtime, with CUDA Toolkit setup, architecture-86 build flags, focused SenseVoice target packaging, docs, and workflow regression tests. Manual workflow dispatch 29656771776 passed all build matrix jobs including build-windows-x64-cuda.

.github/workflows/build-llamacpp-binaries.yml

@@ -82,13 +82,48 @@ jobs:
               -DGGML_FMA=ON
               -DGGML_F16C=ON
               -DGGML_BMI2=ON
+          - os: windows-latest
+            name: windows-x64-cuda
+            cuda: true
+            cuda_version: '13.2.0'
+            cuda_architectures: '86'
+            build_target: llama-funasr-sensevoice
+            timeout_minutes: 90
+            cmake_flags: >-
+              -DGGML_NATIVE=OFF
+              -DGGML_CUDA=ON
+              -DGGML_CUDA_FORCE_CUBLAS=ON
+              -DGGML_CUDA_FA=OFF
+              -DGGML_CUDA_NCCL=OFF
+              -DGGML_CUDA_GRAPHS=OFF
+              -DGGML_AVX=OFF
+              -DGGML_AVX2=OFF
+              -DGGML_AVX_VNNI=OFF
+              -DGGML_AVX512=OFF
+              -DGGML_AVX512_VBMI=OFF
+              -DGGML_AVX512_VNNI=OFF
+              -DGGML_AVX512_BF16=OFF
+              -DGGML_FMA=OFF
+              -DGGML_F16C=OFF
+              -DGGML_BMI2=OFF
     runs-on: ${{ matrix.os }}
+    timeout-minutes: ${{ matrix.timeout_minutes || 30 }}
     defaults:
       run:
         shell: bash
         working-directory: runtime/llama.cpp
     steps:
       - uses: actions/checkout@v4
+      - name: Install CUDA Toolkit
+        if: matrix.cuda
+        uses: Jimver/[email protected]
+        with:
+          cuda: ${{ matrix.cuda_version }}
+          method: network
+          log-file-suffix: ${{ matrix.name }}.txt
+      - name: Show CUDA compiler
+        if: matrix.cuda
+        run: nvcc --version
       - name: Configure and build
         run: |
           # Release binaries must be portable across user machines, not tuned to
@@ -96,8 +131,23 @@ jobs:
           # AVX-VNNI instructions and then crash with SIGILL on ordinary CPUs.
           # Keep the default x64 package conservative, and publish explicit
           # x64-avx2 assets for machines that support AVX2/FMA/F16C/BMI2.
-          cmake -B build -DCMAKE_BUILD_TYPE=Release ${{ matrix.cmake_flags }}
-          cmake --build build --config Release -j 2
+          # The CUDA release asset is SenseVoiceSmall-specific for now; building
+          # only that target keeps the Windows CUDA job bounded and matches the
+          # backend currently exposed by `--backend cuda`.
+          cmake_args=(
+            -B build
+            -DCMAKE_BUILD_TYPE=Release
+            ${{ matrix.cmake_flags }}
+          )
+          if [ -n "${{ matrix.cuda_architectures }}" ]; then
+            cmake_args+=("-DCMAKE_CUDA_ARCHITECTURES=${{ matrix.cuda_architectures }}")
+          fi
+          cmake "${cmake_args[@]}"
+          if [ -n "${{ matrix.build_target }}" ]; then
+            cmake --build build --config Release -j 2 --target "${{ matrix.build_target }}"
+          else
+            cmake --build build --config Release -j 2
+          fi
       - name: Package
         run: |
           mkdir -p pkg
@@ -135,4 +185,4 @@ jobs:
           gh release create "$tag" dist/* \
             --repo "${{ github.repository }}" \
             --title "FunASR llama.cpp runtime $version" \
-            --notes "Prebuilt self-contained binaries for the FunASR llama.cpp / GGUF runtime: SenseVoice, Paraformer and Fun-ASR-Nano with built-in FSMN-VAD. Download the default quantized model with \`bash download-funasr-model.sh <sensevoice|paraformer|nano>\` (the helper requires the Hugging Face CLI: \`pip install -U huggingface_hub\`), then run \`llama-funasr-cli\` / \`llama-funasr-sensevoice\` / \`llama-funasr-paraformer\`. Use the default x64 asset for maximum CPU compatibility; use the x64-avx2 asset on CPUs with AVX2/FMA/F16C/BMI2 for higher throughput. No Python ASR runtime or local build is required. Docs: $docs"
+            --notes "Prebuilt self-contained binaries for the FunASR llama.cpp / GGUF runtime: SenseVoice, Paraformer and Fun-ASR-Nano with built-in FSMN-VAD. Download the default quantized model with \`bash download-funasr-model.sh <sensevoice|paraformer|nano>\` (the helper requires the Hugging Face CLI: \`pip install -U huggingface_hub\`), then run \`llama-funasr-cli\` / \`llama-funasr-sensevoice\` / \`llama-funasr-paraformer\`. Use the default x64 asset for maximum CPU compatibility; use the x64-avx2 asset on CPUs with AVX2/FMA/F16C/BMI2 for higher throughput. The Windows CUDA asset is \`windows-x64-cuda\`; it requires an NVIDIA driver compatible with the CUDA Toolkit version configured by the release workflow, targets CUDA architecture 86, and enables SenseVoiceSmall graph execution with \`llama-funasr-sensevoice ... --backend cuda\`. Build from source for other GPU architectures. No Python ASR runtime or local build is required. Docs: $docs"