[AUTOGENERATED] rocm7.1_internal_testing_IFU_2025-09-08

.bc-linter.yml

@@ -0,0 +1,15 @@
+version: 1
+paths:
+include:
+  - "**/*.py"
+exclude:
+  - ".*"
+  - ".*/**"
+  - "**/.*/**"
+  - "**/.*"
+  - "**/_*/**"
+  - "**/_*.py"
+  - "**/test/**"
+  - "**/benchmarks/**"
+  - "**/test_*.py"
+  - "**/*_test.py"

.ci/aarch64_linux/aarch64_ci_build.sh

@@ -7,6 +7,15 @@ if [[ "$GPU_ARCH_VERSION" == *"12.9"* ]]; then
     export TORCH_CUDA_ARCH_LIST="8.0;9.0;10.0;12.0"
 fi
 
+if [[ "$GPU_ARCH_VERSION" == *"13.0"* ]]; then
+    export TORCH_CUDA_ARCH_LIST="8.0;9.0;10.0;11.0;12.0"
+fi
+
+# Compress the fatbin with -compress-mode=size for CUDA 13
+if [[ "$DESIRED_CUDA" == *"13"* ]]; then
+    export TORCH_NVCC_FLAGS="-compress-mode=size"
+fi
+
 SCRIPTPATH="$( cd -- "$(dirname "$0")" >/dev/null 2>&1 ; pwd -P )"
 source $SCRIPTPATH/aarch64_ci_setup.sh
 

.ci/aarch64_linux/aarch64_wheel_ci_build.py

@@ -77,44 +77,66 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
     wheelname = os.path.basename(wheel_path)
     os.mkdir(f"{folder}/tmp")
     os.system(f"unzip {wheel_path} -d {folder}/tmp")
-    libs_to_copy = [
-        "/usr/local/cuda/extras/CUPTI/lib64/libcupti.so.12",
+    # Common libraries for all CUDA versions
+    common_libs = [
+        # Non-NVIDIA system libraries
+        "/lib64/libgomp.so.1",
+        "/usr/lib64/libgfortran.so.5",
+        "/acl/build/libarm_compute.so",
+        "/acl/build/libarm_compute_graph.so",
+        # Common CUDA libraries (same for all versions)
+        "/usr/local/lib/libnvpl_lapack_lp64_gomp.so.0",
+        "/usr/local/lib/libnvpl_blas_lp64_gomp.so.0",
+        "/usr/local/lib/libnvpl_lapack_core.so.0",
+        "/usr/local/lib/libnvpl_blas_core.so.0",
         "/usr/local/cuda/extras/CUPTI/lib64/libnvperf_host.so",
         "/usr/local/cuda/lib64/libcudnn.so.9",
-        "/usr/local/cuda/lib64/libcublas.so.12",
-        "/usr/local/cuda/lib64/libcublasLt.so.12",
-        "/usr/local/cuda/lib64/libcudart.so.12",
-        "/usr/local/cuda/lib64/libcufft.so.11",
-        "/usr/local/cuda/lib64/libcusparse.so.12",
         "/usr/local/cuda/lib64/libcusparseLt.so.0",
-        "/usr/local/cuda/lib64/libcusolver.so.11",
         "/usr/local/cuda/lib64/libcurand.so.10",
         "/usr/local/cuda/lib64/libnccl.so.2",
-        "/usr/local/cuda/lib64/libnvJitLink.so.12",
-        "/usr/local/cuda/lib64/libnvrtc.so.12",
+        "/usr/local/cuda/lib64/libnvshmem_host.so.3",
         "/usr/local/cuda/lib64/libcudnn_adv.so.9",
         "/usr/local/cuda/lib64/libcudnn_cnn.so.9",
         "/usr/local/cuda/lib64/libcudnn_graph.so.9",
         "/usr/local/cuda/lib64/libcudnn_ops.so.9",
         "/usr/local/cuda/lib64/libcudnn_engines_runtime_compiled.so.9",
         "/usr/local/cuda/lib64/libcudnn_engines_precompiled.so.9",
         "/usr/local/cuda/lib64/libcudnn_heuristic.so.9",
-        "/lib64/libgomp.so.1",
-        "/usr/lib64/libgfortran.so.5",
-        "/acl/build/libarm_compute.so",
-        "/acl/build/libarm_compute_graph.so",
-        "/usr/local/lib/libnvpl_lapack_lp64_gomp.so.0",
-        "/usr/local/lib/libnvpl_blas_lp64_gomp.so.0",
-        "/usr/local/lib/libnvpl_lapack_core.so.0",
-        "/usr/local/lib/libnvpl_blas_core.so.0",
+        "/usr/local/cuda/lib64/libcufile.so.0",
+        "/usr/local/cuda/lib64/libcufile_rdma.so.1",
+        "/usr/local/cuda/lib64/libcusparse.so.12",
     ]
 
-    if "129" in desired_cuda:
-        libs_to_copy += [
-            "/usr/local/cuda/lib64/libnvrtc-builtins.so.12.9",
-            "/usr/local/cuda/lib64/libcufile.so.0",
-            "/usr/local/cuda/lib64/libcufile_rdma.so.1",
+    # CUDA version-specific libraries
+    if "130" in desired_cuda:
+        version_specific_libs = [
+            "/usr/local/cuda/extras/CUPTI/lib64/libcupti.so.13",
+            "/usr/local/cuda/lib64/libcublas.so.13",
+            "/usr/local/cuda/lib64/libcublasLt.so.13",
+            "/usr/local/cuda/lib64/libcudart.so.13",
+            "/usr/local/cuda/lib64/libcufft.so.12",
+            "/usr/local/cuda/lib64/libcusolver.so.12",
+            "/usr/local/cuda/lib64/libnvJitLink.so.13",
+            "/usr/local/cuda/lib64/libnvrtc.so.13",
+            "/usr/local/cuda/lib64/libnvrtc-builtins.so.13.0",
         ]
+    elif "12" in desired_cuda:
+        # Get the last character for libnvrtc-builtins version (e.g., "129" -> "9")
+        minor_version = desired_cuda[-1]
+        version_specific_libs = [
+            "/usr/local/cuda/extras/CUPTI/lib64/libcupti.so.12",
+            "/usr/local/cuda/lib64/libcublas.so.12",
+            "/usr/local/cuda/lib64/libcublasLt.so.12",
+            "/usr/local/cuda/lib64/libcudart.so.12",
+            "/usr/local/cuda/lib64/libcufft.so.11",
+            "/usr/local/cuda/lib64/libcusolver.so.11",
+            "/usr/local/cuda/lib64/libnvJitLink.so.12",
+            "/usr/local/cuda/lib64/libnvrtc.so.12",
+            f"/usr/local/cuda/lib64/libnvrtc-builtins.so.12.{minor_version}",
+        ]
+
+    # Combine all libraries
+    libs_to_copy = common_libs + version_specific_libs
 
     # Copy libraries to unzipped_folder/a/lib
     for lib_path in libs_to_copy:
@@ -208,7 +230,7 @@ def parse_arguments():
     build_vars = "CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000 "
     # MAX_JOB=5 is not required for CPU backend (see commit 465d98b)
     if enable_cuda:
-        build_vars = "MAX_JOBS=5 " + build_vars
+        build_vars += "MAX_JOBS=5 "
 
     override_package_version = os.getenv("OVERRIDE_PACKAGE_VERSION")
     desired_cuda = os.getenv("DESIRED_CUDA")

.ci/aarch64_linux/build_aarch64_wheel.py

@@ -438,9 +438,7 @@ def build_torchvision(
         )
         build_vars += f"BUILD_VERSION={version}.dev{build_date}"
     elif build_version is not None:
-        build_vars += (
-            f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-')[0]}"
-        )
+        build_vars += f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-', maxsplit=1)[0]}"
     if host.using_docker():
         build_vars += " CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000"
 
@@ -495,9 +493,7 @@ def build_torchdata(
         )
         build_vars += f"BUILD_VERSION={version}.dev{build_date}"
     elif build_version is not None:
-        build_vars += (
-            f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-')[0]}"
-        )
+        build_vars += f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-', maxsplit=1)[0]}"
     if host.using_docker():
         build_vars += " CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000"
 
@@ -553,9 +549,7 @@ def build_torchtext(
         )
         build_vars += f"BUILD_VERSION={version}.dev{build_date}"
     elif build_version is not None:
-        build_vars += (
-            f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-')[0]}"
-        )
+        build_vars += f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-', maxsplit=1)[0]}"
     if host.using_docker():
         build_vars += " CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000"
 
@@ -613,9 +607,7 @@ def build_torchaudio(
         )
         build_vars += f"BUILD_VERSION={version}.dev{build_date}"
     elif build_version is not None:
-        build_vars += (
-            f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-')[0]}"
-        )
+        build_vars += f"BUILD_VERSION={build_version} PYTORCH_VERSION={branch[1:].split('-', maxsplit=1)[0]}"
     if host.using_docker():
         build_vars += " CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000"
 

.ci/docker/README.md

@@ -120,8 +120,8 @@ If your new Docker image needs a library installed from a specific pinned commit
    If you're introducing a new argument to the Docker build, make sure to add it in the Docker build step in `.ci/docker/build.sh`:
    ```bash
    docker build \
-      ....
-      --build-arg "NEW_ARG_1=${NEW_ARG_1}"
+     ....
+     --build-arg "NEW_ARG_1=${NEW_ARG_1}"
    ```
 
 3. **Update Dockerfile logic**:

.ci/docker/almalinux/Dockerfile

@@ -64,6 +64,10 @@ FROM cuda as cuda12.9
 RUN bash ./install_cuda.sh 12.9
 ENV DESIRED_CUDA=12.9
 
+FROM cuda as cuda13.0
+RUN bash ./install_cuda.sh 13.0
+ENV DESIRED_CUDA=13.0
+
 FROM ${ROCM_IMAGE} as rocm
 ENV PYTORCH_ROCM_ARCH="gfx900;gfx906;gfx908;gfx90a;gfx942;gfx1030;gfx1100;gfx1101;gfx1102;gfx1200;gfx1201"
 ADD ./common/install_mkl.sh install_mkl.sh
@@ -76,10 +80,10 @@ ADD ./common/install_mnist.sh install_mnist.sh
 RUN bash ./install_mnist.sh
 
 FROM base as all_cuda
-COPY --from=cuda11.8  /usr/local/cuda-11.8 /usr/local/cuda-11.8
 COPY --from=cuda12.6  /usr/local/cuda-12.6 /usr/local/cuda-12.6
 COPY --from=cuda12.8  /usr/local/cuda-12.8 /usr/local/cuda-12.8
 COPY --from=cuda12.9  /usr/local/cuda-12.9 /usr/local/cuda-12.9
+COPY --from=cuda13.0  /usr/local/cuda-13.0 /usr/local/cuda-13.0
 
 # Final step
 FROM ${BASE_TARGET} as final