ci: build cuda 12.4 for scalellm cpp images (#346)

vectorch-ai · Oct 15, 2024 · 2550fba · 2550fba
1 parent 8e0320b
commit 2550fba
Show file tree

Hide file tree

Showing 3 changed files with 48 additions and 6 deletions.
diff --git a/.github/workflows/publish_cpp_image.yml b/.github/workflows/publish_cpp_image.yml
@@ -136,3 +136,47 @@ jobs:
             vectorchai/scalellm_cu118:${{ inputs.tag }}
             vectorchai/scalellm_cu118:latest
 
+  publish_scalellm_cuda124:
+    runs-on: [self-hosted, linux, release]
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - name: Build ScaleLLM for cuda 12.4
+        timeout-minutes: 60
+        run: |
+          docker pull vectorchai/scalellm_devel:cuda12.4
+          docker run --rm -t \
+            -v "$CI_CACHE_DIR":/ci_cache \
+            -v "$GITHUB_WORKSPACE":/ScaleLLM \
+            -e VCPKG_DEFAULT_BINARY_CACHE=/ci_cache/.vcpkg/bincache \
+            -e CCACHE_DIR=/ci_cache/.ccache \
+            -u $(id -u):$(id -g) \
+            vectorchai/scalellm_devel:cuda12.4 \
+            bash /ScaleLLM/scripts/build_scalellm.sh
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKER_HUB_USER }}
+          password: ${{ secrets.DOCKER_HUB_TOKEN }}
+
+      - name: Build and push scalellm for cuda 12.4
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          file: ./docker/Dockerfile.scalellm
+          push: true
+          no-cache: true
+          tags: |
+            vectorchai/scalellm_cu124:${{ inputs.tag }}
+            vectorchai/scalellm_cu124:latest
+
diff --git a/.github/workflows/publish_devel_image.yml b/.github/workflows/publish_devel_image.yml
@@ -7,7 +7,7 @@ env:
 
 jobs:
   publish_base:
-    runs-on: [self-hosted, linux, release]
+    runs-on: [self-hosted, linux, build]
     steps:
       - name: Checkout repository
         uses: actions/checkout@v4

diff --git a/.github/workflows/publish_manylinux_image.yml b/.github/workflows/publish_manylinux_image.yml
@@ -11,9 +11,7 @@ jobs:
       fail-fast: false
       matrix:
         cuda: ["11.8", "12.1", "12.4"]
-    runs-on: [self-hosted, linux, release]
-    env:
-      CUDA_VERSION: ${{ matrix.cuda }}
+    runs-on: [self-hosted, linux, build]
     steps:
       - name: Checkout repository
         uses: actions/checkout@v4
@@ -42,7 +40,7 @@ jobs:
           cache-from: type=local,src=$CI_CACHE_DIR/.buildx-cache
           cache-to: type=local,dest=$CI_CACHE_DIR/.buildx-cache
           build-args: |
-            CUDA_VERSION=${CUDA_VERSION}
+            CUDA_VERSION=${{ matrix.cuda }}
           tags: |
-            vectorchai/scalellm_manylinux:cuda${CUDA_VERSION}
+            vectorchai/scalellm_manylinux:cuda${{ matrix.cuda }}