fix cuda 13.0 cuda-arch-list to be 8.9 for L4 gpu test

.ci/docker/build.sh

@@ -116,7 +116,7 @@ case "$tag" in
     INSTALL_MINGW=yes
     ;;
   pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc11)
-    CUDA_VERSION=13.0.0
+    CUDA_VERSION=13.0.2
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     VISION=yes
@@ -125,6 +125,16 @@ case "$tag" in
     UCC_COMMIT=${_UCC_COMMIT}
     TRITON=yes
     ;;
+  pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc9)
+    CUDA_VERSION=13.0.2
+    ANACONDA_PYTHON_VERSION=3.10
+    GCC_VERSION=9
+    VISION=yes
+    KATEX=yes
+    UCX_COMMIT=${_UCX_COMMIT}
+    UCC_COMMIT=${_UCC_COMMIT}
+    TRITON=yes
+    ;;
   pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc9-inductor-benchmarks)
     CUDA_VERSION=12.8.1
     ANACONDA_PYTHON_VERSION=3.10

.github/workflows/docker-builds.yml

@@ -54,6 +54,7 @@ jobs:
           pytorch-linux-jammy-cuda12.8-cudnn9-py3.12-gcc11-vllm,
           pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc9-inductor-benchmarks,
           pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc9,
+          pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc9,
           pytorch-linux-jammy-cuda12.4-cudnn9-py3-gcc11,
           pytorch-linux-jammy-py3.10-clang12,
           pytorch-linux-jammy-py3.11-clang12,

.github/workflows/periodic.yml

@@ -204,6 +204,39 @@ jobs:
       test-matrix: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc11-build.outputs.test-matrix }}
     secrets: inherit
 
+  linux-jammy-cuda13_0-py3_10-gcc9-debug-build:
+    name: linux-jammy-cuda13.0-py3.10-gcc9-debug
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc9-debug
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc9
+      cuda-arch-list: 8.6
+      test-matrix: |
+        { include: [
+          { config: "default", shard: 1, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 2, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 3, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 4, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 5, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 6, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+          { config: "default", shard: 7, num_shards: 7, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu", owners: ["oncall:debug-build"] },
+        ]}
+    secrets: inherit
+
+  linux-jammy-cuda13_0-py3_10-gcc9-debug-test:
+    name: linux-jammy-cuda13.0-py3.10-gcc9-debug
+    uses: ./.github/workflows/_linux-test.yml
+    needs:
+      - linux-jammy-cuda13_0-py3_10-gcc9-debug-build
+      - target-determination
+    with:
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc9-debug
+      docker-image: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc9-debug-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc9-debug-build.outputs.test-matrix }}
+    secrets: inherit
+
   linux-jammy-cuda12_8-py3-gcc11-slow-gradcheck-build:
     name: linux-jammy-cuda12.8-py3-gcc11-slow-gradcheck
     uses: ./.github/workflows/_linux-build.yml

.github/workflows/pull.yml

@@ -268,6 +268,35 @@ jobs:
         ]}
     secrets: inherit
 
+  linux-jammy-cuda13_0-py3_10-gcc9-build:
+    name: linux-jammy-cuda13.0-py3.10-gcc9
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc9
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc9
+      cuda-arch-list: 8.9
+      test-matrix: |
+        { include: [
+          { config: "default", shard: 1, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 2, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 3, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 4, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 5, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+        ]}
+    secrets: inherit
+
+  linux-jammy-cuda13_0-py3_10-gcc9-test:
+    name: linux-jammy-cuda13.0-py3.10-gcc9
+    uses: ./.github/workflows/_linux-test.yml
+    needs: linux-jammy-cuda13_0-py3_10-gcc9-build
+    with:
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc9
+      docker-image: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc9-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc9-build.outputs.test-matrix }}
+    secrets: inherit
+
   linux-jammy-cpu-py3_10-gcc11-bazel-test:
     name: linux-jammy-cpu-py3.10-gcc11-bazel-test
     uses: ./.github/workflows/_bazel-build-test.yml

.github/workflows/slow.yml

@@ -78,6 +78,35 @@ jobs:
       test-matrix: ${{ needs.linux-jammy-cuda12_8-py3_10-gcc11-sm86-build.outputs.test-matrix }}
     secrets: inherit
 
+  linux-jammy-cuda13_0-py3_10-gcc11-sm86-build:
+    name: linux-jammy-cuda13.0-py3.10-gcc11-sm86
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc11-sm86
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc11
+      cuda-arch-list: 8.6
+      test-matrix: |
+        { include: [
+          { config: "slow", shard: 1, num_shards: 3, runner: "linux.g5.4xlarge.nvidia.gpu" },
+          { config: "slow", shard: 2, num_shards: 3, runner: "linux.g5.4xlarge.nvidia.gpu" },
+          { config: "slow", shard: 3, num_shards: 3, runner: "linux.g5.4xlarge.nvidia.gpu" },
+        ]}
+    secrets: inherit
+
+  linux-jammy-cuda13_0-py3_10-gcc11-sm86-test:
+    name: linux-jammy-cuda13.0-py3.10-gcc11-sm86
+    uses: ./.github/workflows/_linux-test.yml
+    needs:
+      - linux-jammy-cuda13_0-py3_10-gcc11-sm86-build
+      - target-determination
+    with:
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc11-sm86
+      docker-image: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc11-sm86-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc11-sm86-build.outputs.test-matrix }}
+    secrets: inherit
+
   linux-jammy-py3_10-clang12-build:
     name: linux-jammy-py3.10-clang12
     uses: ./.github/workflows/_linux-build.yml

.github/workflows/trunk.yml

@@ -63,6 +63,23 @@ jobs:
         ]}
     secrets: inherit
 
+  libtorch-linux-jammy-cuda13_0-py3_10-gcc11-debug-build:
+    name: libtorch-linux-jammy-cuda13.0-py3.10-gcc11-debug
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      build-environment: libtorch-linux-jammy-cuda13.0-py3.10-gcc11
+      cuda-arch-list: '7.5 8.9'
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc11
+      build-generates-artifacts: false
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      runner: "linux.c7i.4xlarge"
+      test-matrix: |
+        { include: [
+          { config: "default", shard: 1, num_shards: 1 },
+        ]}
+    secrets: inherit
+
   linux-jammy-cuda12_8-py3_10-gcc11-build:
     name: linux-jammy-cuda12.8-py3.10-gcc11
     uses: ./.github/workflows/_linux-build.yml
@@ -99,6 +116,41 @@ jobs:
       test-matrix: ${{ needs.linux-jammy-cuda12_8-py3_10-gcc11-build.outputs.test-matrix }}
     secrets: inherit
 
+  linux-jammy-cuda13_0-py3_10-gcc11-build:
+    name: linux-jammy-cuda13.0-py3.10-gcc11
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc11
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc11
+      cuda-arch-list: '7.5 8.9'
+      test-matrix: |
+        { include: [
+          { config: "default", shard: 1, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 2, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 3, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 4, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "default", shard: 5, num_shards: 5, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g6.4xlarge.experimental.nvidia.gpu" },
+          { config: "distributed", shard: 1, num_shards: 3, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g4dn.12xlarge.nvidia.gpu" },
+          { config: "distributed", shard: 2, num_shards: 3, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g4dn.12xlarge.nvidia.gpu" },
+          { config: "distributed", shard: 3, num_shards: 3, runner: "${{ needs.get-label-type.outputs.label-type }}linux.g4dn.12xlarge.nvidia.gpu" },
+          { config: "pr_time_benchmarks", shard: 1, num_shards: 1, runner: "linux.g4dn.metal.nvidia.gpu" },
+        ]}
+    secrets: inherit
+
+  linux-jammy-cuda13_0-py3_10-gcc11-test:
+    name: linux-jammy-cuda13.0-py3.10-gcc11
+    uses: ./.github/workflows/_linux-test.yml
+    needs:
+      - linux-jammy-cuda13_0-py3_10-gcc11-build
+      - target-determination
+    with:
+      timeout-minutes: 360
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc11
+      docker-image: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc11-build.outputs.docker-image }}
+      test-matrix: ${{ needs.linux-jammy-cuda13_0-py3_10-gcc11-build.outputs.test-matrix }}
+    secrets: inherit
 
   # no-ops builds test USE_PER_OPERATOR_HEADERS=0 where ATen/ops is not generated
   linux-jammy-cuda12_8-py3_10-gcc11-no-ops-build:
@@ -115,6 +167,21 @@ jobs:
         ]}
     secrets: inherit
 
+  linux-jammy-cuda13_0-py3_10-gcc11-no-ops-build:
+    name: linux-jammy-cuda13.0-py3.10-gcc11-no-ops
+    uses: ./.github/workflows/_linux-build.yml
+    needs: get-label-type
+    with:
+      runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
+      build-environment: linux-jammy-cuda13.0-py3.10-gcc11-no-ops
+      cuda-arch-list: '7.5 8.9'
+      docker-image-name: ci-image:pytorch-linux-jammy-cuda13.0-cudnn9-py3-gcc11
+      test-matrix: |
+        { include: [
+          { config: "default", shard: 1, num_shards: 1 },
+        ]}
+    secrets: inherit
+
   macos-py3-arm64-build:
     if: github.repository_owner == 'pytorch'
     name: macos-py3-arm64

cmake/Dependencies.cmake

@@ -1394,6 +1394,9 @@ if(NOT INTERN_BUILD_MOBILE)
   # https://github.com/pytorch/pytorch/pull/55292
   string(APPEND CMAKE_CUDA_FLAGS " -DCUB_WRAPPED_NAMESPACE=at_cuda_detail")
 
+  # Suppress cusparse warnings
+  string(APPEND CMAKE_CUDA_FLAGS " -DDISABLE_CUSPARSE_DEPRECATED")
+
   message(STATUS "Found CUDA with FP16 support, compiling with torch.cuda.HalfTensor")
   string(APPEND CMAKE_CUDA_FLAGS " -DCUDA_HAS_FP16=1"
                                  " -D__CUDA_NO_HALF_OPERATORS__"