Halves time spent in generating the key strings

.ci/aarch64_linux/aarch64_ci_build.sh

@@ -27,7 +27,6 @@ if [ "$DESIRED_CUDA" = "cpu" ]; then
     USE_PRIORITIZED_TEXT_FOR_LD=1 python /pytorch/.ci/aarch64_linux/aarch64_wheel_ci_build.py --enable-mkldnn
 else
     echo "BASE_CUDA_VERSION is set to: $DESIRED_CUDA"
-    export USE_SYSTEM_NCCL=1
     #USE_PRIORITIZED_TEXT_FOR_LD for enable linker script optimization https://github.com/pytorch/pytorch/pull/121975/files
     USE_PRIORITIZED_TEXT_FOR_LD=1 python /pytorch/.ci/aarch64_linux/aarch64_wheel_ci_build.py --enable-mkldnn --enable-cuda
 fi

.ci/aarch64_linux/aarch64_wheel_ci_build.py

@@ -31,47 +31,33 @@ def build_ArmComputeLibrary() -> None:
         "build=native",
     ]
     acl_install_dir = "/acl"
-    acl_checkout_dir = os.getenv("ACL_SOURCE_DIR", "ComputeLibrary")
-    if os.path.isdir(acl_install_dir):
-        shutil.rmtree(acl_install_dir)
-    if not os.path.isdir(acl_checkout_dir) or not len(os.listdir(acl_checkout_dir)):
-        check_call(
-            [
-                "git",
-                "clone",
-                "https://github.com/ARM-software/ComputeLibrary.git",
-                "-b",
-                "v25.02",
-                "--depth",
-                "1",
-                "--shallow-submodules",
-            ]
-        )
+    acl_checkout_dir = "ComputeLibrary"
+    os.makedirs(acl_install_dir)
+    check_call(
+        [
+            "git",
+            "clone",
+            "https://github.com/ARM-software/ComputeLibrary.git",
+            "-b",
+            "v25.02",
+            "--depth",
+            "1",
+            "--shallow-submodules",
+        ]
+    )
 
     check_call(
-        ["scons", "Werror=1", f"-j{os.cpu_count()}"] + acl_build_flags,
+        ["scons", "Werror=1", "-j8", f"build_dir=/{acl_install_dir}/build"]
+        + acl_build_flags,
         cwd=acl_checkout_dir,
     )
-    for d in ["arm_compute", "include", "utils", "support", "src", "build"]:
+    for d in ["arm_compute", "include", "utils", "support", "src"]:
         shutil.copytree(f"{acl_checkout_dir}/{d}", f"{acl_install_dir}/{d}")
 
 
-def replace_tag(filename) -> None:
-    with open(filename) as f:
-        lines = f.readlines()
-    for i, line in enumerate(lines):
-        if line.startswith("Tag:"):
-            lines[i] = line.replace("-linux_", "-manylinux_2_28_")
-            print(f"Updated tag from {line} to {lines[i]}")
-            break
-
-    with open(filename, "w") as f:
-        f.writelines(lines)
-
-
-def package_cuda_wheel(wheel_path, desired_cuda) -> None:
+def update_wheel(wheel_path, desired_cuda) -> None:
     """
-    Package the cuda wheel libraries
+    Update the cuda wheel libraries
     """
     folder = os.path.dirname(wheel_path)
     wheelname = os.path.basename(wheel_path)
@@ -102,19 +88,30 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
         "/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",
     ]
-
-    if "128" in desired_cuda:
+    if enable_cuda:
         libs_to_copy += [
-            "/usr/local/cuda/lib64/libnvrtc-builtins.so.12.8",
-            "/usr/local/cuda/lib64/libcufile.so.0",
-            "/usr/local/cuda/lib64/libcufile_rdma.so.1",
+            "/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",
+        ]
+        if "126" in desired_cuda:
+            libs_to_copy += [
+                "/usr/local/cuda/lib64/libnvrtc-builtins.so.12.6",
+                "/usr/local/cuda/lib64/libcufile.so.0",
+                "/usr/local/cuda/lib64/libcufile_rdma.so.1",
+            ]
+        elif "128" in desired_cuda:
+            libs_to_copy += [
+                "/usr/local/cuda/lib64/libnvrtc-builtins.so.12.8",
+                "/usr/local/cuda/lib64/libcufile.so.0",
+                "/usr/local/cuda/lib64/libcufile_rdma.so.1",
+            ]
+    else:
+        libs_to_copy += [
+            "/opt/OpenBLAS/lib/libopenblas.so.0",
         ]
-
     # Copy libraries to unzipped_folder/a/lib
     for lib_path in libs_to_copy:
         lib_name = os.path.basename(lib_path)
@@ -123,13 +120,6 @@ def package_cuda_wheel(wheel_path, desired_cuda) -> None:
             f"cd {folder}/tmp/torch/lib/; "
             f"patchelf --set-rpath '$ORIGIN' --force-rpath {folder}/tmp/torch/lib/{lib_name}"
         )
-
-    # Make sure the wheel is tagged with manylinux_2_28
-    for f in os.scandir(f"{folder}/tmp/"):
-        if f.is_dir() and f.name.endswith(".dist-info"):
-            replace_tag(f"{f.path}/WHEEL")
-            break
-
     os.mkdir(f"{folder}/cuda_wheel")
     os.system(f"cd {folder}/tmp/; zip -r {folder}/cuda_wheel/{wheelname} *")
     shutil.move(
@@ -204,10 +194,8 @@ if __name__ == "__main__":
     ).decode()
 
     print("Building PyTorch wheel")
-    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 CMAKE_SHARED_LINKER_FLAGS=-Wl,-z,max-page-size=0x10000 "
+    os.system("cd /pytorch; python setup.py clean")
 
     override_package_version = os.getenv("OVERRIDE_PACKAGE_VERSION")
     desired_cuda = os.getenv("DESIRED_CUDA")
@@ -254,6 +242,6 @@ if __name__ == "__main__":
         print("Updating Cuda Dependency")
         filename = os.listdir("/pytorch/dist/")
         wheel_path = f"/pytorch/dist/{filename[0]}"
-        package_cuda_wheel(wheel_path, desired_cuda)
+        update_wheel(wheel_path, desired_cuda)
     pytorch_wheel_name = complete_wheel("/pytorch/")
     print(f"Build Complete. Created {pytorch_wheel_name}..")

.ci/caffe2/README.md

@@ -10,3 +10,5 @@ example: `py2-cuda9.0-cudnn7-ubuntu16.04`. The Docker images that are
 built on Jenkins and are used in triggered builds already have this
 environment variable set in their manifest. Also see
 `./docker/jenkins/*/Dockerfile` and search for `BUILD_ENVIRONMENT`.
+
+Our Jenkins installation is located at https://ci.pytorch.org/jenkins/.

.ci/caffe2/test.sh

@@ -13,6 +13,10 @@ if [[ "${BUILD_ENVIRONMENT}" == *-android* ]]; then
   echo 'Skipping tests'
   exit 0
 fi
+if [[ "${BUILD_ENVIRONMENT}" == *-rocm* ]]; then
+  # temporary to locate some kernel issues on the CI nodes
+  export HSAKMT_DEBUG_LEVEL=4
+fi
 # These additional packages are needed for circleci ROCm builds.
 if [[ $BUILD_ENVIRONMENT == *rocm* ]]; then
     # Need networkx 2.0 because bellmand_ford was moved in 2.1 . Scikit-image by

.ci/docker/README.md

@@ -34,5 +34,5 @@ See `build.sh` for valid build environments (it's the giant switch).
 ./build.sh pytorch-linux-bionic-py3.8-gcc9 -t myimage:latest
 
 # Set flags (see build.sh) and build image
-sudo bash -c 'TRITON=1 ./build.sh pytorch-linux-bionic-py3.8-gcc9 -t myimage:latest
+sudo bash -c 'PROTOBUF=1 ./build.sh pytorch-linux-bionic-py3.8-gcc9 -t myimage:latest
 ```

.ci/docker/almalinux/Dockerfile

@@ -1,7 +1,6 @@
 ARG CUDA_VERSION=12.4
 ARG BASE_TARGET=cuda${CUDA_VERSION}
-ARG ROCM_IMAGE=rocm/dev-almalinux-8:6.3-complete
-FROM amd64/almalinux:8.10-20250519 as base
+FROM amd64/almalinux:8 as base
 
 ENV LC_ALL en_US.UTF-8
 ENV LANG en_US.UTF-8
@@ -9,10 +8,12 @@ ENV LANGUAGE en_US.UTF-8
 
 ARG DEVTOOLSET_VERSION=11
 
+ENV LC_ALL en_US.UTF-8
+ENV LANG en_US.UTF-8
+ENV LANGUAGE en_US.UTF-8
+
 RUN yum -y update
 RUN yum -y install epel-release
-# install glibc-langpack-en make sure en_US.UTF-8 locale is available
-RUN yum -y install glibc-langpack-en
 RUN yum install -y sudo wget curl perl util-linux xz bzip2 git patch which perl zlib-devel openssl-devel yum-utils autoconf automake make gcc-toolset-${DEVTOOLSET_VERSION}-toolchain
 # Just add everything as a safe.directory for git since these will be used in multiple places with git
 RUN git config --global --add safe.directory '*'
@@ -40,12 +41,11 @@ RUN bash ./install_conda.sh && rm install_conda.sh
 
 # Install CUDA
 FROM base as cuda
-ARG CUDA_VERSION=12.6
+ARG CUDA_VERSION=12.4
 RUN rm -rf /usr/local/cuda-*
 ADD ./common/install_cuda.sh install_cuda.sh
 COPY ./common/install_nccl.sh install_nccl.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
 ENV CUDA_HOME=/usr/local/cuda-${CUDA_VERSION}
 # Preserve CUDA_VERSION for the builds
 ENV CUDA_VERSION=${CUDA_VERSION}
@@ -56,20 +56,18 @@ FROM cuda as cuda11.8
 RUN bash ./install_cuda.sh 11.8
 ENV DESIRED_CUDA=11.8
 
+FROM cuda as cuda12.1
+RUN bash ./install_cuda.sh 12.1
+ENV DESIRED_CUDA=12.1
+
+FROM cuda as cuda12.4
+RUN bash ./install_cuda.sh 12.4
+ENV DESIRED_CUDA=12.4
+
 FROM cuda as cuda12.6
 RUN bash ./install_cuda.sh 12.6
 ENV DESIRED_CUDA=12.6
 
-FROM cuda as cuda12.8
-RUN bash ./install_cuda.sh 12.8
-ENV DESIRED_CUDA=12.8
-
-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
-RUN bash ./install_mkl.sh && rm install_mkl.sh
-ENV MKLROOT /opt/intel
-
 # Install MNIST test data
 FROM base as mnist
 ADD ./common/install_mnist.sh install_mnist.sh
@@ -77,8 +75,9 @@ 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=cuda12.1  /usr/local/cuda-12.1 /usr/local/cuda-12.1
+COPY --from=cuda12.4  /usr/local/cuda-12.4 /usr/local/cuda-12.4
 COPY --from=cuda12.6  /usr/local/cuda-12.6 /usr/local/cuda-12.6
-COPY --from=cuda12.4  /usr/local/cuda-12.8 /usr/local/cuda-12.8
 
 # Final step
 FROM ${BASE_TARGET} as final

.ci/docker/almalinux/build.sh

@@ -15,16 +15,9 @@ fi
 DOCKER_TAG_PREFIX=$(echo "${image}" | awk -F':' '{print $2}')
 
 CUDA_VERSION=""
-ROCM_VERSION=""
-EXTRA_BUILD_ARGS=""
 if [[ "${DOCKER_TAG_PREFIX}" == cuda* ]]; then
     # extract cuda version from image name and tag.  e.g. manylinux2_28-builder:cuda12.8 returns 12.8
     CUDA_VERSION=$(echo "${DOCKER_TAG_PREFIX}" | awk -F'cuda' '{print $2}')
-    EXTRA_BUILD_ARGS="--build-arg CUDA_VERSION=${CUDA_VERSION}"
-elif [[ "${DOCKER_TAG_PREFIX}" == rocm* ]]; then
-    # extract rocm version from image name and tag.  e.g. manylinux2_28-builder:rocm6.2.4 returns 6.2.4
-    ROCM_VERSION=$(echo "${DOCKER_TAG_PREFIX}" | awk -F'rocm' '{print $2}')
-    EXTRA_BUILD_ARGS="--build-arg ROCM_IMAGE=rocm/dev-almalinux-8:${ROCM_VERSION}-complete"
 fi
 
 case ${DOCKER_TAG_PREFIX} in
@@ -34,9 +27,6 @@ case ${DOCKER_TAG_PREFIX} in
   cuda*)
     BASE_TARGET=cuda${CUDA_VERSION}
     ;;
-  rocm*)
-    BASE_TARGET=rocm
-    ;;
   *)
     echo "ERROR: Unknown docker tag ${DOCKER_TAG_PREFIX}"
     exit 1
@@ -57,8 +47,8 @@ docker build \
   --target final \
   --progress plain \
   --build-arg "BASE_TARGET=${BASE_TARGET}" \
+  --build-arg "CUDA_VERSION=${CUDA_VERSION}" \
   --build-arg "DEVTOOLSET_VERSION=11" \
-  ${EXTRA_BUILD_ARGS} \
   -t ${tmp_tag} \
   $@ \
   -f "${TOPDIR}/.ci/docker/almalinux/Dockerfile" \

.ci/docker/build.sh

@@ -85,6 +85,9 @@ elif [[ "$image" == *linter* ]]; then
   DOCKERFILE="linter/Dockerfile"
 fi
 
+# CMake 3.18 is needed to support CUDA17 language variant
+CMAKE_VERSION=3.18.5
+
 _UCX_COMMIT=7bb2722ff2187a0cad557ae4a6afa090569f83fb
 _UCC_COMMIT=20eae37090a4ce1b32bcce6144ccad0b49943e0b
 if [[ "$image" == *rocm* ]]; then
@@ -92,32 +95,62 @@ if [[ "$image" == *rocm* ]]; then
   _UCC_COMMIT=0c0fc21559835044ab107199e334f7157d6a0d3d
 fi
 
-tag=$(echo $image | awk -F':' '{print $2}')
-
 # It's annoying to rename jobs every time you want to rewrite a
 # configuration, so we hardcode everything here rather than do it
 # from scratch
-case "$tag" in
+case "$image" in
   pytorch-linux-focal-cuda12.6-cudnn9-py3-gcc11)
     CUDA_VERSION=12.6.3
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
-  pytorch-linux-jammy-cuda12.8-cudnn9-py3-gcc9-inductor-benchmarks)
-    CUDA_VERSION=12.8
+  pytorch-linux-focal-cuda12.4-cudnn9-py3-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.4.1
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
+    TRITON=yes
+    INDUCTOR_BENCHMARKS=yes
+    ;;
+  pytorch-linux-focal-cuda12.4-cudnn9-py3.12-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.4.1
+    CUDNN_VERSION=9
+    ANACONDA_PYTHON_VERSION=3.12
+    GCC_VERSION=9
+    PROTOBUF=yes
+    VISION=yes
+    KATEX=yes
+    UCX_COMMIT=${_UCX_COMMIT}
+    UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
+    TRITON=yes
+    INDUCTOR_BENCHMARKS=yes
+    ;;
+  pytorch-linux-focal-cuda12.4-cudnn9-py3.13-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.4.1
+    CUDNN_VERSION=9
+    ANACONDA_PYTHON_VERSION=3.13
+    GCC_VERSION=9
+    PROTOBUF=yes
+    VISION=yes
+    KATEX=yes
+    UCX_COMMIT=${_UCX_COMMIT}
+    UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     INDUCTOR_BENCHMARKS=yes
     ;;
@@ -126,45 +159,53 @@ case "$tag" in
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
-  pytorch-linux-jammy-cuda12.6-cudnn9-py3-gcc9-inductor-benchmarks)
-    CUDA_VERSION=12.6
+  pytorch-linux-focal-cuda12.6-cudnn9-py3-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.6.3
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     INDUCTOR_BENCHMARKS=yes
     ;;
-  pytorch-linux-jammy-cuda12.6-cudnn9-py3.12-gcc9-inductor-benchmarks)
-    CUDA_VERSION=12.6
+  pytorch-linux-focal-cuda12.6-cudnn9-py3.12-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.6.3
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.12
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     INDUCTOR_BENCHMARKS=yes
     ;;
-  pytorch-linux-jammy-cuda12.6-cudnn9-py3.13-gcc9-inductor-benchmarks)
-    CUDA_VERSION=12.6
+  pytorch-linux-focal-cuda12.6-cudnn9-py3.13-gcc9-inductor-benchmarks)
+    CUDA_VERSION=12.6.3
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.13
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     INDUCTOR_BENCHMARKS=yes
     ;;
@@ -173,81 +214,101 @@ case "$tag" in
     CUDNN_VERSION=9
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
   pytorch-linux-focal-py3-clang10-onnx)
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=10
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     ONNX=yes
     ;;
   pytorch-linux-focal-py3.9-clang10)
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=10
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
   pytorch-linux-focal-py3.11-clang10)
     ANACONDA_PYTHON_VERSION=3.11
     CLANG_VERSION=10
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
   pytorch-linux-focal-py3.9-gcc9)
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=9
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
-  pytorch-linux-jammy-rocm-n-1-py3)
+  pytorch-linux-focal-rocm-n-1-py3)
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
+    PROTOBUF=yes
+    VISION=yes
+    ROCM_VERSION=6.2.4
+    NINJA_VERSION=1.9.0
+    CONDA_CMAKE=yes
+    TRITON=yes
+    KATEX=yes
+    UCX_COMMIT=${_UCX_COMMIT}
+    UCC_COMMIT=${_UCC_COMMIT}
+    INDUCTOR_BENCHMARKS=yes
+    ;;
+  pytorch-linux-focal-rocm-n-py3)
+    ANACONDA_PYTHON_VERSION=3.10
+    GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
     ROCM_VERSION=6.3
     NINJA_VERSION=1.9.0
+    CONDA_CMAKE=yes
     TRITON=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
     UCC_COMMIT=${_UCC_COMMIT}
     INDUCTOR_BENCHMARKS=yes
     ;;
-  pytorch-linux-jammy-rocm-n-py3)
-    ANACONDA_PYTHON_VERSION=3.10
+  pytorch-linux-jammy-xpu-2024.0-py3)
+    ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
-    ROCM_VERSION=6.4
+    XPU_VERSION=0.5
     NINJA_VERSION=1.9.0
+    CONDA_CMAKE=yes
     TRITON=yes
-    KATEX=yes
-    UCX_COMMIT=${_UCX_COMMIT}
-    UCC_COMMIT=${_UCC_COMMIT}
-    INDUCTOR_BENCHMARKS=yes
     ;;
   pytorch-linux-jammy-xpu-2025.0-py3)
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
     XPU_VERSION=2025.0
     NINJA_VERSION=1.9.0
-    TRITON=yes
-    ;;
-  pytorch-linux-jammy-xpu-2025.1-py3)
-    ANACONDA_PYTHON_VERSION=3.9
-    GCC_VERSION=11
-    VISION=yes
-    XPU_VERSION=2025.1
-    NINJA_VERSION=1.9.0
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
     pytorch-linux-jammy-py3.9-gcc11-inductor-benchmarks)
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     DOCS=yes
     INDUCTOR_BENCHMARKS=yes
@@ -257,30 +318,37 @@ case "$tag" in
     CUDA_VERSION=11.8
     CUDNN_VERSION=9
     CLANG_VERSION=12
+    PROTOBUF=yes
     VISION=yes
     TRITON=yes
     ;;
   pytorch-linux-jammy-py3-clang12-asan)
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=12
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     ;;
   pytorch-linux-jammy-py3-clang15-asan)
     ANACONDA_PYTHON_VERSION=3.10
     CLANG_VERSION=15
+    CONDA_CMAKE=yes
     VISION=yes
     ;;
   pytorch-linux-jammy-py3-clang18-asan)
     ANACONDA_PYTHON_VERSION=3.10
     CLANG_VERSION=18
+    CONDA_CMAKE=yes
     VISION=yes
     ;;
   pytorch-linux-jammy-py3.9-gcc11)
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
+    PROTOBUF=yes
     VISION=yes
     KATEX=yes
+    CONDA_CMAKE=yes
     TRITON=yes
     DOCS=yes
     UNINSTALL_DILL=yes
@@ -288,12 +356,14 @@ case "$tag" in
   pytorch-linux-jammy-py3-clang12-executorch)
     ANACONDA_PYTHON_VERSION=3.10
     CLANG_VERSION=12
+    CONDA_CMAKE=yes
     EXECUTORCH=yes
     ;;
   pytorch-linux-jammy-py3.12-halide)
     CUDA_VERSION=12.6
     ANACONDA_PYTHON_VERSION=3.12
     GCC_VERSION=11
+    CONDA_CMAKE=yes
     HALIDE=yes
     TRITON=yes
     ;;
@@ -301,6 +371,7 @@ case "$tag" in
     CUDA_VERSION=12.6
     ANACONDA_PYTHON_VERSION=3.12
     GCC_VERSION=11
+    CONDA_CMAKE=yes
     TRITON_CPU=yes
     ;;
   pytorch-linux-focal-linter)
@@ -308,16 +379,20 @@ case "$tag" in
     # We will need to update mypy version eventually, but that's for another day. The task
     # would be to upgrade mypy to 1.0.0 with Python 3.11
     PYTHON_VERSION=3.9
+    PIP_CMAKE=yes
     ;;
   pytorch-linux-jammy-cuda11.8-cudnn9-py3.9-linter)
     PYTHON_VERSION=3.9
     CUDA_VERSION=11.8
+    PIP_CMAKE=yes
     ;;
   pytorch-linux-jammy-aarch64-py3.10-gcc11)
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     ACL=yes
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     # snadampal: skipping llvm src build install because the current version
     # from pytorch/llvm:9.0.1 is x86 specific
     SKIP_LLVM_SRC_BUILD_INSTALL=yes
@@ -326,7 +401,9 @@ case "$tag" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     ACL=yes
+    PROTOBUF=yes
     VISION=yes
+    CONDA_CMAKE=yes
     # snadampal: skipping llvm src build install because the current version
     # from pytorch/llvm:9.0.1 is x86 specific
     SKIP_LLVM_SRC_BUILD_INSTALL=yes
@@ -334,6 +411,7 @@ case "$tag" in
     ;;
   *)
     # Catch-all for builds that are not hardcoded.
+    PROTOBUF=yes
     VISION=yes
     echo "image '$image' did not match an existing build configuration"
     if [[ "$image" == *py* ]]; then
@@ -349,7 +427,8 @@ case "$tag" in
       TRITON=yes
       # To ensure that any ROCm config will build using conda cmake
       # and thus have LAPACK/MKL enabled
-      fi
+      CONDA_CMAKE=yes
+    fi
     if [[ "$image" == *centos7* ]]; then
       NINJA_VERSION=1.10.2
     fi
@@ -365,6 +444,9 @@ case "$tag" in
     if [[ "$image" == *glibc* ]]; then
       extract_version_from_image_name glibc GLIBC_VERSION
     fi
+    if [[ "$image" == *cmake* ]]; then
+      extract_version_from_image_name cmake CMAKE_VERSION
+    fi
   ;;
 esac
 
@@ -391,6 +473,7 @@ docker build \
        ${no_cache_flag} \
        ${progress_flag} \
        --build-arg "BUILD_ENVIRONMENT=${image}" \
+       --build-arg "PROTOBUF=${PROTOBUF:-}" \
        --build-arg "LLVMDEV=${LLVMDEV:-}" \
        --build-arg "VISION=${VISION:-}" \
        --build-arg "UBUNTU_VERSION=${UBUNTU_VERSION}" \
@@ -405,6 +488,7 @@ docker build \
        --build-arg "CUDNN_VERSION=${CUDNN_VERSION}" \
        --build-arg "TENSORRT_VERSION=${TENSORRT_VERSION}" \
        --build-arg "GRADLE_VERSION=${GRADLE_VERSION}" \
+       --build-arg "CMAKE_VERSION=${CMAKE_VERSION:-}" \
        --build-arg "NINJA_VERSION=${NINJA_VERSION:-}" \
        --build-arg "KATEX=${KATEX:-}" \
        --build-arg "ROCM_VERSION=${ROCM_VERSION:-}" \
@@ -412,6 +496,8 @@ docker build \
        --build-arg "IMAGE_NAME=${IMAGE_NAME}" \
        --build-arg "UCX_COMMIT=${UCX_COMMIT}" \
        --build-arg "UCC_COMMIT=${UCC_COMMIT}" \
+       --build-arg "CONDA_CMAKE=${CONDA_CMAKE}" \
+       --build-arg "PIP_CMAKE=${PIP_CMAKE}" \
        --build-arg "TRITON=${TRITON}" \
        --build-arg "TRITON_CPU=${TRITON_CPU}" \
        --build-arg "ONNX=${ONNX}" \
@@ -420,7 +506,6 @@ docker build \
        --build-arg "EXECUTORCH=${EXECUTORCH}" \
        --build-arg "HALIDE=${HALIDE}" \
        --build-arg "XPU_VERSION=${XPU_VERSION}" \
-       --build-arg "UNINSTALL_DILL=${UNINSTALL_DILL}" \
        --build-arg "ACL=${ACL:-}" \
        --build-arg "SKIP_SCCACHE_INSTALL=${SKIP_SCCACHE_INSTALL:-}" \
        --build-arg "SKIP_LLVM_SRC_BUILD_INSTALL=${SKIP_LLVM_SRC_BUILD_INSTALL:-}" \
@@ -497,12 +582,3 @@ elif [ "$HAS_TRITON" = "yes" ]; then
   echo "expecting triton to not be installed, but it is"
   exit 1
 fi
-
-# Sanity check cmake version.  Executorch reinstalls cmake and I'm not sure if
-# they support 4.0.0 yet, so exclude them from this check.
-CMAKE_VERSION=$(drun cmake --version)
-if [[ "$EXECUTORCH" != *yes* && "$CMAKE_VERSION" != *4.* ]]; then
-  echo "CMake version is not 4.0.0:"
-  drun cmake --version
-  exit 1
-fi

.ci/docker/centos-rocm/Dockerfile

@@ -17,8 +17,9 @@ RUN bash ./install_base.sh && rm install_base.sh
 # Update CentOS git version
 RUN yum -y remove git
 RUN yum -y remove git-*
-RUN yum -y install https://packages.endpointdev.com/rhel/7/os/x86_64/endpoint-repo-1.9-1.x86_64.rpm && \
-    sed -i 's/packages.endpoint/packages.endpointdev/' /etc/yum.repos.d/endpoint.repo
+RUN yum -y install https://packages.endpoint.com/rhel/7/os/x86_64/endpoint-repo-1.9-1.x86_64.rpm || \
+    (yum -y install https://packages.endpointdev.com/rhel/7/os/x86_64/endpoint-repo-1.9-1.x86_64.rpm && \
+    sed -i "s/packages.endpoint/packages.endpointdev/" /etc/yum.repos.d/endpoint.repo)
 RUN yum install -y git
 
 # Install devtoolset
@@ -39,6 +40,7 @@ RUN bash ./install_user.sh && rm install_user.sh
 
 # Install conda and other packages (e.g., numpy, pytest)
 ARG ANACONDA_PYTHON_VERSION
+ARG CONDA_CMAKE
 ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
 ENV PATH /opt/conda/envs/py_$ANACONDA_PYTHON_VERSION/bin:/opt/conda/bin:$PATH
 COPY requirements-ci.txt /opt/conda/requirements-ci.txt
@@ -46,6 +48,13 @@ COPY ./common/install_conda.sh install_conda.sh
 COPY ./common/common_utils.sh common_utils.sh
 RUN bash ./install_conda.sh && rm install_conda.sh common_utils.sh /opt/conda/requirements-ci.txt
 
+# (optional) Install protobuf for ONNX
+ARG PROTOBUF
+COPY ./common/install_protobuf.sh install_protobuf.sh
+RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
+RUN rm install_protobuf.sh
+ENV INSTALLED_PROTOBUF ${PROTOBUF}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./
@@ -73,6 +82,12 @@ ENV MAGMA_HOME /opt/rocm/magma
 ENV LANG en_US.utf8
 ENV LC_ALL en_US.utf8
 
+# (optional) Install non-default CMake version
+ARG CMAKE_VERSION
+COPY ./common/install_cmake.sh install_cmake.sh
+RUN if [ -n "${CMAKE_VERSION}" ]; then bash ./install_cmake.sh; fi
+RUN rm install_cmake.sh
+
 # (optional) Install non-default Ninja version
 ARG NINJA_VERSION
 COPY ./common/install_ninja.sh install_ninja.sh

.ci/docker/ci_commit_pins/executorch.txt

@@ -1 +1 @@
-b173722085b3f555d6ba4533d6bbaddfd7c71144
+381ae5d57d35c165d98df728380b20fbde350392

.ci/docker/ci_commit_pins/nccl-cu12.txt

@@ -1 +1 @@
-v2.26.5-1
+v2.26.2-1

.ci/docker/ci_commit_pins/triton-xpu.txt

@@ -1 +1 @@
-b0e26b7359c147b8aa0af686c20510fb9b15990a
+0bcc8265e677e5321606a3311bf71470f14456a8

.ci/docker/ci_commit_pins/triton.txt

@@ -1 +1 @@
-c8757738a7418249896224430ce84888e8ecdd79
+96316ce50fade7e209553aba4898cd9b82aab83b

.ci/docker/common/install_base.sh

@@ -99,6 +99,9 @@ install_centos() {
 
   ccache_deps="asciidoc docbook-dtds docbook-style-xsl libxslt"
   numpy_deps="gcc-gfortran"
+  # Note: protobuf-c-{compiler,devel} on CentOS are too old to be used
+  # for Caffe2. That said, we still install them to make sure the build
+  # system opts to build/use protoc and libprotobuf from third-party.
   yum install -y \
     $ccache_deps \
     $numpy_deps \

.ci/docker/common/install_cache.sh

@@ -9,7 +9,7 @@ install_ubuntu() {
   # Instead use lib and headers from OpenSSL1.1 installed in `install_openssl.sh``
   apt-get install -y cargo
   echo "Checking out sccache repo"
-  git clone https://github.com/mozilla/sccache -b v0.10.0
+  git clone https://github.com/mozilla/sccache -b v0.9.1
   cd sccache
   echo "Building sccache"
   cargo build --release

.ci/docker/common/install_cmake.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+set -ex
+
+[ -n "$CMAKE_VERSION" ]
+
+# Remove system cmake install so it won't get used instead
+ID=$(grep -oP '(?<=^ID=).+' /etc/os-release | tr -d '"')
+case "$ID" in
+  ubuntu)
+    apt-get remove cmake -y
+    ;;
+  centos)
+    yum remove cmake -y
+    ;;
+  *)
+    echo "Unable to determine OS..."
+    exit 1
+    ;;
+esac
+
+# Turn 3.6.3 into v3.6
+path=$(echo "${CMAKE_VERSION}" | sed -e 's/\([0-9].[0-9]\+\).*/v\1/')
+file="cmake-${CMAKE_VERSION}-Linux-x86_64.tar.gz"
+
+# Download and install specific CMake version in /usr/local
+pushd /tmp
+curl -Os --retry 3 "https://cmake.org/files/${path}/${file}"
+tar -C /usr/local --strip-components 1 --no-same-owner -zxf cmake-*.tar.gz
+rm -f cmake-*.tar.gz
+popd

.ci/docker/common/install_conda.sh

@@ -7,7 +7,7 @@ if [ -n "$ANACONDA_PYTHON_VERSION" ]; then
   BASE_URL="https://repo.anaconda.com/miniconda"
   CONDA_FILE="Miniconda3-latest-Linux-x86_64.sh"
   if [[ $(uname -m) == "aarch64" ]] || [[ "$BUILD_ENVIRONMENT" == *xpu* ]]; then
-    BASE_URL="https://github.com/conda-forge/miniforge/releases/latest/download"  # @lint-ignore
+    BASE_URL="https://github.com/conda-forge/miniforge/releases/latest/download"
     CONDA_FILE="Miniforge3-Linux-$(uname -m).sh"
   fi
 
@@ -75,11 +75,19 @@ if [ -n "$ANACONDA_PYTHON_VERSION" ]; then
   # and libpython-static for torch deploy
   conda_install llvmdev=8.0.0 "libpython-static=${ANACONDA_PYTHON_VERSION}"
 
+  # Use conda cmake in some cases. Conda cmake will be newer than our supported
+  # min version (3.5 for xenial and 3.10 for bionic), so we only do it in those
+  # following builds that we know should use conda. Specifically, Ubuntu bionic
+  # and focal cannot find conda mkl with stock cmake, so we need a cmake from conda
+  if [ -n "${CONDA_CMAKE}" ]; then
+    conda_install cmake
+  fi
+
   # Magma package names are concatenation of CUDA major and minor ignoring revision
   # I.e. magma-cuda102 package corresponds to CUDA_VERSION=10.2 and CUDA_VERSION=10.2.89
   # Magma is installed from a tarball in the ossci-linux bucket into the conda env
   if [ -n "$CUDA_VERSION" ]; then
-    conda_run ${SCRIPT_FOLDER}/install_magma_conda.sh $(cut -f1-2 -d'.' <<< ${CUDA_VERSION})
+    ${SCRIPT_FOLDER}/install_magma_conda.sh $(cut -f1-2 -d'.' <<< ${CUDA_VERSION}) ${ANACONDA_PYTHON_VERSION}
   fi
 
   # Install some other packages, including those needed for Python test reporting

.ci/docker/common/install_cpython.sh

@@ -3,7 +3,7 @@
 set -uex -o pipefail
 
 PYTHON_DOWNLOAD_URL=https://www.python.org/ftp/python
-PYTHON_DOWNLOAD_GITHUB_BRANCH=https://github.com/python/cpython/archive/refs/heads  # @lint-ignore
+PYTHON_DOWNLOAD_GITHUB_BRANCH=https://github.com/python/cpython/archive/refs/heads
 GET_PIP_URL=https://bootstrap.pypa.io/get-pip.py
 
 # Python versions to be installed in /opt/$VERSION_NO

.ci/docker/common/install_cuda.sh

@@ -2,54 +2,64 @@
 
 set -ex
 
-arch_path=''
-targetarch=${TARGETARCH:-$(uname -m)}
-if [ ${targetarch} = 'amd64' ] || [ "${targetarch}" = 'x86_64' ]; then
-  arch_path='x86_64'
-else
-  arch_path='sbsa'
-fi
+CUDNN_VERSION=9.5.1.17
 
-function install_cuda {
-  version=$1
-  runfile=$2
-  major_minor=${version%.*}
-  rm -rf /usr/local/cuda-${major_minor} /usr/local/cuda
-  if [[ ${arch_path} == 'sbsa' ]]; then
-      runfile="${runfile}_sbsa"
-  fi
-  runfile="${runfile}.run"
-  wget -q https://developer.download.nvidia.com/compute/cuda/${version}/local_installers/${runfile} -O ${runfile}
-  chmod +x ${runfile}
-  ./${runfile} --toolkit --silent
-  rm -f ${runfile}
-  rm -f /usr/local/cuda && ln -s /usr/local/cuda-${major_minor} /usr/local/cuda
+function install_cusparselt_040 {
+    # cuSparseLt license: https://docs.nvidia.com/cuda/cusparselt/license.html
+    mkdir tmp_cusparselt && pushd tmp_cusparselt
+    wget -q https://developer.download.nvidia.com/compute/cusparselt/redist/libcusparse_lt/linux-x86_64/libcusparse_lt-linux-x86_64-0.4.0.7-archive.tar.xz
+    tar xf libcusparse_lt-linux-x86_64-0.4.0.7-archive.tar.xz
+    cp -a libcusparse_lt-linux-x86_64-0.4.0.7-archive/include/* /usr/local/cuda/include/
+    cp -a libcusparse_lt-linux-x86_64-0.4.0.7-archive/lib/* /usr/local/cuda/lib64/
+    popd
+    rm -rf tmp_cusparselt
 }
 
-function install_cudnn {
-  cuda_major_version=$1
-  cudnn_version=$2
-  mkdir tmp_cudnn && cd tmp_cudnn
-  # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
-  filepath="cudnn-linux-${arch_path}-${cudnn_version}_cuda${cuda_major_version}-archive"
-  wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-${arch_path}/${filepath}.tar.xz
-  tar xf ${filepath}.tar.xz
-  cp -a ${filepath}/include/* /usr/local/cuda/include/
-  cp -a ${filepath}/lib/* /usr/local/cuda/lib64/
-  cd ..
-  rm -rf tmp_cudnn
+function install_cusparselt_062 {
+    # cuSparseLt license: https://docs.nvidia.com/cuda/cusparselt/license.html
+    mkdir tmp_cusparselt && pushd tmp_cusparselt
+    wget -q https://developer.download.nvidia.com/compute/cusparselt/redist/libcusparse_lt/linux-x86_64/libcusparse_lt-linux-x86_64-0.6.2.3-archive.tar.xz
+    tar xf libcusparse_lt-linux-x86_64-0.6.2.3-archive.tar.xz
+    cp -a libcusparse_lt-linux-x86_64-0.6.2.3-archive/include/* /usr/local/cuda/include/
+    cp -a libcusparse_lt-linux-x86_64-0.6.2.3-archive/lib/* /usr/local/cuda/lib64/
+    popd
+    rm -rf tmp_cusparselt
+}
+
+function install_cusparselt_063 {
+    # cuSparseLt license: https://docs.nvidia.com/cuda/cusparselt/license.html
+    mkdir tmp_cusparselt && pushd tmp_cusparselt
+    wget -q https://developer.download.nvidia.com/compute/cusparselt/redist/libcusparse_lt/linux-x86_64/libcusparse_lt-linux-x86_64-0.6.3.2-archive.tar.xz
+    tar xf libcusparse_lt-linux-x86_64-0.6.3.2-archive.tar.xz
+    cp -a libcusparse_lt-linux-x86_64-0.6.3.2-archive/include/* /usr/local/cuda/include/
+    cp -a libcusparse_lt-linux-x86_64-0.6.3.2-archive/lib/* /usr/local/cuda/lib64/
+    popd
+    rm -rf tmp_cusparselt
 }
 
 function install_118 {
     CUDNN_VERSION=9.1.0.70
     echo "Installing CUDA 11.8 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.4.0"
-    install_cuda 11.8.0 cuda_11.8.0_520.61.05_linux
+    rm -rf /usr/local/cuda-11.8 /usr/local/cuda
+    # install CUDA 11.8.0 in the same container
+    wget -q https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_520.61.05_linux.run
+    chmod +x cuda_11.8.0_520.61.05_linux.run
+    ./cuda_11.8.0_520.61.05_linux.run --toolkit --silent
+    rm -f cuda_11.8.0_520.61.05_linux.run
+    rm -f /usr/local/cuda && ln -s /usr/local/cuda-11.8 /usr/local/cuda
 
-    install_cudnn 11 $CUDNN_VERSION
+    # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
+    mkdir tmp_cudnn && cd tmp_cudnn
+    wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/cudnn-linux-x86_64-${CUDNN_VERSION}_cuda11-archive.tar.xz -O cudnn-linux-x86_64-${CUDNN_VERSION}_cuda11-archive.tar.xz
+    tar xf cudnn-linux-x86_64-${CUDNN_VERSION}_cuda11-archive.tar.xz
+    cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda11-archive/include/* /usr/local/cuda/include/
+    cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda11-archive/lib/* /usr/local/cuda/lib64/
+    cd ..
+    rm -rf tmp_cudnn
 
     CUDA_VERSION=11.8 bash install_nccl.sh
 
-    CUDA_VERSION=11.8 bash install_cusparselt.sh
+    install_cusparselt_040
 
     ldconfig
 }
@@ -57,27 +67,52 @@ function install_118 {
 function install_124 {
   CUDNN_VERSION=9.1.0.70
   echo "Installing CUDA 12.4.1 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.6.2"
-  install_cuda 12.4.1 cuda_12.4.1_550.54.15_linux
+  rm -rf /usr/local/cuda-12.4 /usr/local/cuda
+  # install CUDA 12.4.1 in the same container
+  wget -q https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda_12.4.1_550.54.15_linux.run
+  chmod +x cuda_12.4.1_550.54.15_linux.run
+  ./cuda_12.4.1_550.54.15_linux.run --toolkit --silent
+  rm -f cuda_12.4.1_550.54.15_linux.run
+  rm -f /usr/local/cuda && ln -s /usr/local/cuda-12.4 /usr/local/cuda
 
-  install_cudnn 12 $CUDNN_VERSION
+  # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
+  mkdir tmp_cudnn && cd tmp_cudnn
+  wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz -O cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  tar xf cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/include/* /usr/local/cuda/include/
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/lib/* /usr/local/cuda/lib64/
+  cd ..
+  rm -rf tmp_cudnn
 
   CUDA_VERSION=12.4 bash install_nccl.sh
 
-  CUDA_VERSION=12.4 bash install_cusparselt.sh
+  install_cusparselt_062
 
   ldconfig
 }
 
 function install_126 {
-  CUDNN_VERSION=9.5.1.17
   echo "Installing CUDA 12.6.3 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.6.3"
-  install_cuda 12.6.3 cuda_12.6.3_560.35.05_linux
+  rm -rf /usr/local/cuda-12.6 /usr/local/cuda
+  # install CUDA 12.6.3 in the same container
+  wget -q https://developer.download.nvidia.com/compute/cuda/12.6.3/local_installers/cuda_12.6.3_560.35.05_linux.run
+  chmod +x cuda_12.6.3_560.35.05_linux.run
+  ./cuda_12.6.3_560.35.05_linux.run --toolkit --silent
+  rm -f cuda_12.6.3_560.35.05_linux.run
+  rm -f /usr/local/cuda && ln -s /usr/local/cuda-12.6 /usr/local/cuda
 
-  install_cudnn 12 $CUDNN_VERSION
+  # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
+  mkdir tmp_cudnn && cd tmp_cudnn
+  wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz -O cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  tar xf cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/include/* /usr/local/cuda/include/
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/lib/* /usr/local/cuda/lib64/
+  cd ..
+  rm -rf tmp_cudnn
 
   CUDA_VERSION=12.6 bash install_nccl.sh
 
-  CUDA_VERSION=12.6 bash install_cusparselt.sh
+  install_cusparselt_063
 
   ldconfig
 }
@@ -183,16 +218,27 @@ function prune_126 {
 
 function install_128 {
   CUDNN_VERSION=9.8.0.87
-  echo "Installing CUDA 12.8.1 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.6.3"
-  # install CUDA 12.8.1 in the same container
-  install_cuda 12.8.1 cuda_12.8.1_570.124.06_linux
+  echo "Installing CUDA 12.8.0 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.6.3"
+  rm -rf /usr/local/cuda-12.8 /usr/local/cuda
+  # install CUDA 12.8.0 in the same container
+  wget -q https://developer.download.nvidia.com/compute/cuda/12.8.0/local_installers/cuda_12.8.0_570.86.10_linux.run
+  chmod +x cuda_12.8.0_570.86.10_linux.run
+  ./cuda_12.8.0_570.86.10_linux.run --toolkit --silent
+  rm -f cuda_12.8.0_570.86.10_linux.run
+  rm -f /usr/local/cuda && ln -s /usr/local/cuda-12.8 /usr/local/cuda
 
   # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
-  install_cudnn 12 $CUDNN_VERSION
+  mkdir tmp_cudnn && cd tmp_cudnn
+  wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-x86_64/cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz -O cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  tar xf cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/include/* /usr/local/cuda/include/
+  cp -a cudnn-linux-x86_64-${CUDNN_VERSION}_cuda12-archive/lib/* /usr/local/cuda/lib64/
+  cd ..
+  rm -rf tmp_cudnn
 
   CUDA_VERSION=12.8 bash install_nccl.sh
 
-  CUDA_VERSION=12.8 bash install_cusparselt.sh
+  install_cusparselt_063
 
   ldconfig
 }

.ci/docker/common/install_cuda_aarch64.sh

@@ -0,0 +1,55 @@
+#!/bin/bash
+# Script used only in CD pipeline
+
+set -ex
+
+CUDNN_VERSION=9.8.0.87
+
+function install_cusparselt_063 {
+    # cuSparseLt license: https://docs.nvidia.com/cuda/cusparselt/license.html
+    mkdir tmp_cusparselt && pushd tmp_cusparselt
+    wget -q https://developer.download.nvidia.com/compute/cusparselt/redist/libcusparse_lt/linux-sbsa/libcusparse_lt-linux-sbsa-0.6.3.2-archive.tar.xz
+    tar xf libcusparse_lt-linux-sbsa-0.6.3.2-archive.tar.xz
+    cp -a libcusparse_lt-linux-sbsa-0.6.3.2-archive/include/* /usr/local/cuda/include/
+    cp -a libcusparse_lt-linux-sbsa-0.6.3.2-archive/lib/* /usr/local/cuda/lib64/
+    popd
+    rm -rf tmp_cusparselt
+}
+
+function install_128 {
+  echo "Installing CUDA 12.8.0 and cuDNN ${CUDNN_VERSION} and NCCL and cuSparseLt-0.6.3"
+  rm -rf /usr/local/cuda-12.8 /usr/local/cuda
+  # install CUDA 12.8.0 in the same container
+  wget -q https://developer.download.nvidia.com/compute/cuda/12.8.0/local_installers/cuda_12.8.0_570.86.10_linux_sbsa.run
+  chmod +x cuda_12.8.0_570.86.10_linux_sbsa.run
+  ./cuda_12.8.0_570.86.10_linux_sbsa.run --toolkit --silent
+  rm -f cuda_12.8.0_570.86.10_linux_sbsa.run
+  rm -f /usr/local/cuda && ln -s /usr/local/cuda-12.8 /usr/local/cuda
+
+  # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
+  mkdir tmp_cudnn && cd tmp_cudnn
+  wget -q https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/linux-sbsa/cudnn-linux-sbsa-${CUDNN_VERSION}_cuda12-archive.tar.xz -O cudnn-linux-sbsa-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  tar xf cudnn-linux-sbsa-${CUDNN_VERSION}_cuda12-archive.tar.xz
+  cp -a cudnn-linux-sbsa-${CUDNN_VERSION}_cuda12-archive/include/* /usr/local/cuda/include/
+  cp -a cudnn-linux-sbsa-${CUDNN_VERSION}_cuda12-archive/lib/* /usr/local/cuda/lib64/
+  cd ..
+  rm -rf tmp_cudnn
+
+  CUDA_VERSION=12.8 bash install_nccl.sh
+
+  install_cusparselt_063
+
+  ldconfig
+}
+
+# idiomatic parameter and option handling in sh
+while test $# -gt 0
+do
+    case "$1" in
+    12.8) install_128;
+        ;;
+    *) echo "bad argument $1"; exit 1
+        ;;
+    esac
+    shift
+done

.ci/docker/common/install_executorch.sh

@@ -13,7 +13,7 @@ clone_executorch() {
   # and fetch the target commit
   pushd executorch
   git checkout "${EXECUTORCH_PINNED_COMMIT}"
-  git submodule update --init --recursive
+  git submodule update --init
   popd
 
   chown -R jenkins executorch

.ci/docker/common/install_halide.sh

@@ -17,7 +17,7 @@ if [ -n "${UBUNTU_VERSION}" ];then
                   libopenblas-dev libeigen3-dev libatlas-base-dev libzstd-dev
 fi
 
-pip_install numpy scipy imageio cmake ninja
+conda_install numpy scipy imageio cmake ninja
 
 git clone --depth 1 --branch release/16.x --recursive https://github.com/llvm/llvm-project.git
 cmake -DCMAKE_BUILD_TYPE=Release \

.ci/docker/common/install_inductor_benchmark_deps.sh

@@ -14,9 +14,16 @@ function install_timm() {
   local commit
   commit=$(get_pinned_commit timm)
 
+  # TODO (huydhn): There is no torchvision release on 3.13 when I write this, so
+  # I'm using nightly here instead. We just need to package to be able to install
+  # TIMM. Removing this once vision has a release on 3.13
+  if [[ "${ANACONDA_PYTHON_VERSION}" == "3.13" ]]; then
+    pip_install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/cu124
+  fi
+
   pip_install "git+https://github.com/huggingface/pytorch-image-models@${commit}"
   # Clean up
-  conda_run pip uninstall -y torch torchvision triton
+  conda_run pip uninstall -y cmake torch torchvision triton
 }
 
 # Pango is needed for weasyprint which is needed for doctr

.ci/docker/common/install_magma_conda.sh

@@ -1,23 +1,26 @@
 #!/usr/bin/env bash
-# Script that installs magma from tarball inside conda environment.
-# It replaces anaconda magma-cuda package which is no longer published.
-# Execute it inside active conda environment.
-# See issue: https://github.com/pytorch/pytorch/issues/138506
+# Script that replaces the magma install from a conda package
 
 set -eou pipefail
 
-cuda_version_nodot=${1/./}
-anaconda_dir=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
+function do_install() {
+    cuda_version_nodot=${1/./}
+    anaconda_python_version=$2
 
-MAGMA_VERSION="2.6.1"
-magma_archive="magma-cuda${cuda_version_nodot}-${MAGMA_VERSION}-1.tar.bz2"
-(
-    set -x
-    tmp_dir=$(mktemp -d)
-    pushd ${tmp_dir}
-    curl -OLs https://ossci-linux.s3.us-east-1.amazonaws.com/${magma_archive}
-    tar -xvf "${magma_archive}"
-    mv include/* "${anaconda_dir}/include/"
-    mv lib/* "${anaconda_dir}/lib"
-    popd
-)
+    MAGMA_VERSION="2.6.1"
+    magma_archive="magma-cuda${cuda_version_nodot}-${MAGMA_VERSION}-1.tar.bz2"
+
+    anaconda_dir="/opt/conda/envs/py_${anaconda_python_version}"
+    (
+        set -x
+        tmp_dir=$(mktemp -d)
+        pushd ${tmp_dir}
+        curl -OLs https://ossci-linux.s3.us-east-1.amazonaws.com/${magma_archive}
+        tar -xvf "${magma_archive}"
+        mv include/* "${anaconda_dir}/include/"
+        mv lib/* "${anaconda_dir}/lib"
+        popd
+    )
+}
+
+do_install $1 $2

.ci/docker/common/install_onnx.sh

@@ -8,6 +8,16 @@ retry () {
     "$@" || (sleep 10 && "$@") || (sleep 20 && "$@") || (sleep 40 && "$@")
 }
 
+# A bunch of custom pip dependencies for ONNX
+pip_install \
+  beartype==0.15.0 \
+  filelock==3.9.0 \
+  flatbuffers==2.0 \
+  mock==5.0.1 \
+  ninja==1.10.2 \
+  networkx==2.5 \
+  numpy==1.24.2
+
 # ONNXRuntime should be installed before installing
 # onnx-weekly. Otherwise, onnx-weekly could be
 # overwritten by onnx.
@@ -19,8 +29,12 @@ pip_install \
   transformers==4.36.2
 
 pip_install coloredlogs packaging
+
 pip_install onnxruntime==1.18.1
-pip_install onnxscript==0.3.0
+pip_install onnx==1.17.0
+pip_install onnxscript==0.2.2 --no-deps
+# required by onnxscript
+pip_install ml_dtypes
 
 # Cache the transformers model to be used later by ONNX tests. We need to run the transformers
 # package to download the model. By default, the model is cached at ~/.cache/huggingface/hub/

.ci/docker/common/install_protobuf.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+
+set -ex
+
+pb_dir="/usr/temp_pb_install_dir"
+mkdir -p $pb_dir
+
+# On the nvidia/cuda:9-cudnn7-devel-centos7 image we need this symlink or
+# else it will fail with
+#   g++: error: ./../lib64/crti.o: No such file or directory
+ln -s /usr/lib64 "$pb_dir/lib64"
+
+curl -LO "https://github.com/protocolbuffers/protobuf/releases/download/v3.17.3/protobuf-all-3.17.3.tar.gz" --retry 3
+
+tar -xvz --no-same-owner -C "$pb_dir" --strip-components 1 -f protobuf-all-3.17.3.tar.gz
+NPROC=$[$(nproc) - 2]
+pushd "$pb_dir" && ./configure && make -j${NPROC} && make -j${NPROC} check && sudo make -j${NRPOC} install && sudo ldconfig
+popd
+rm -rf $pb_dir

.ci/docker/common/install_python.sh

@@ -13,3 +13,6 @@ source /var/lib/jenkins/ci_env/bin/activate
 
 python -mpip install --upgrade pip
 python -mpip install -r /opt/requirements-ci.txt
+if [ -n "${PIP_CMAKE}" ]; then
+  python -mpip install cmake==3.31.6
+fi

.ci/docker/common/install_rocm.sh

@@ -66,25 +66,17 @@ EOF
     done
 
     # ROCm 6.3 had a regression where initializing static code objects had significant overhead
-    # ROCm 6.4 did not yet fix the regression, also HIP branch names are different
-    if [[ $(ver $ROCM_VERSION) -eq $(ver 6.3) ]] || [[ $(ver $ROCM_VERSION) -eq $(ver 6.4) ]]; then
-        if [[ $(ver $ROCM_VERSION) -eq $(ver 6.3) ]]; then
-            HIP_BRANCH=rocm-6.3.x
-            VER_STR=6.3
-        elif [[ $(ver $ROCM_VERSION) -eq $(ver 6.4) ]]; then
-            HIP_BRANCH=release/rocm-rel-6.4
-            VER_STR=6.4
-        fi
+    if [[ $(ver $ROCM_VERSION) -eq $(ver 6.3) ]]; then
         # clr build needs CppHeaderParser but can only find it using conda's python
         /opt/conda/bin/python -m pip install CppHeaderParser
-        git clone https://github.com/ROCm/HIP -b $HIP_BRANCH
+        git clone https://github.com/ROCm/HIP -b rocm-6.3.x
         HIP_COMMON_DIR=$(readlink -f HIP)
-        git clone https://github.com/jeffdaily/clr -b release/rocm-rel-${VER_STR}-statco-hotfix
+        git clone https://github.com/jeffdaily/clr -b release/rocm-rel-6.3-statco-hotfix
         mkdir -p clr/build
         pushd clr/build
         cmake .. -DCLR_BUILD_HIP=ON -DHIP_COMMON_DIR=$HIP_COMMON_DIR
         make -j
-        cp hipamd/lib/libamdhip64.so.${VER_STR}.* /opt/rocm/lib/libamdhip64.so.${VER_STR}.*
+        cp hipamd/lib/libamdhip64.so.6.3.* /opt/rocm/lib/libamdhip64.so.6.3.*
         popd
         rm -rf HIP clr
     fi

.ci/docker/common/install_triton.sh

@@ -10,8 +10,12 @@ fi
 
 source "$(dirname "${BASH_SOURCE[0]}")/common_utils.sh"
 
-get_pip_version() {
-  conda_run pip list | grep -w $* | head -n 1 | awk '{print $2}'
+get_conda_version() {
+  as_jenkins conda list -n py_$ANACONDA_PYTHON_VERSION | grep -w $* | head -n 1 | awk '{print $2}'
+}
+
+conda_reinstall() {
+  as_jenkins conda install -q -n py_$ANACONDA_PYTHON_VERSION -y --force-reinstall $*
 }
 
 if [ -n "${XPU_VERSION}" ]; then
@@ -33,9 +37,11 @@ if [ -n "${UBUNTU_VERSION}" ];then
     apt-get install -y gpg-agent
 fi
 
-# Keep the current cmake and numpy version here, so we can reinstall them later
-CMAKE_VERSION=$(get_pip_version cmake)
-NUMPY_VERSION=$(get_pip_version numpy)
+if [ -n "${CONDA_CMAKE}" ]; then
+  # Keep the current cmake and numpy version here, so we can reinstall them later
+  CMAKE_VERSION=$(get_conda_version cmake)
+  NUMPY_VERSION=$(get_conda_version numpy)
+fi
 
 if [ -z "${MAX_JOBS}" ]; then
     export MAX_JOBS=$(nproc)
@@ -51,12 +57,7 @@ as_jenkins git clone --recursive ${TRITON_REPO} triton
 cd triton
 as_jenkins git checkout ${TRITON_PINNED_COMMIT}
 as_jenkins git submodule update --init --recursive
-
-# Old versions of python have setup.py in ./python; newer versions have it in ./
-if [ ! -f setup.py ]; then
-  cd python
-fi
-
+cd python
 pip_install pybind11==2.13.6
 
 # TODO: remove patch setup.py once we have a proper fix for https://github.com/triton-lang/triton/issues/4527
@@ -82,19 +83,17 @@ cp dist/*.whl /opt/triton
 # Install the wheel for docker builds that don't use multi stage
 pip_install dist/*.whl
 
-# TODO: This is to make sure that the same cmake and numpy version from install conda
-# script is used. Without this step, the newer cmake version (3.25.2) downloaded by
-# triton build step via pip will fail to detect conda MKL. Once that issue is fixed,
-# this can be removed.
-#
-# The correct numpy version also needs to be set here because conda claims that it
-# causes inconsistent environment.  Without this, conda will attempt to install the
-# latest numpy version, which fails ASAN tests with the following import error: Numba
-# needs NumPy 1.20 or less.
-# Note that we install numpy with pip as conda might not have the version we want
-if [ -n "${CMAKE_VERSION}" ]; then
-  pip_install "cmake==${CMAKE_VERSION}"
-fi
-if [ -n "${NUMPY_VERSION}" ]; then
-  pip_install "numpy==${NUMPY_VERSION}"
+if [ -n "${CONDA_CMAKE}" ]; then
+  # TODO: This is to make sure that the same cmake and numpy version from install conda
+  # script is used. Without this step, the newer cmake version (3.25.2) downloaded by
+  # triton build step via pip will fail to detect conda MKL. Once that issue is fixed,
+  # this can be removed.
+  #
+  # The correct numpy version also needs to be set here because conda claims that it
+  # causes inconsistent environment.  Without this, conda will attempt to install the
+  # latest numpy version, which fails ASAN tests with the following import error: Numba
+  # needs NumPy 1.20 or less.
+  conda_reinstall cmake="${CMAKE_VERSION}"
+  # Note that we install numpy with pip as conda might not have the version we want
+  pip_install --force-reinstall numpy=="${NUMPY_VERSION}"
 fi

.ci/docker/common/install_xpu.sh

@@ -26,7 +26,7 @@ function install_ubuntu() {
     wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB \
         | gpg --dearmor > /usr/share/keyrings/oneapi-archive-keyring.gpg.gpg
     echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg.gpg] \
-        https://apt.repos.intel.com/oneapi all main" \
+        https://apt.repos.intel.com/${XPU_REPO_NAME} all main" \
         | tee /etc/apt/sources.list.d/oneAPI.list
 
     # Update the packages list and repository index
@@ -74,7 +74,7 @@ function install_rhel() {
     tee > /etc/yum.repos.d/oneAPI.repo << EOF
 [oneAPI]
 name=Intel for Pytorch GPU dev repository
-baseurl=https://yum.repos.intel.com/oneapi
+baseurl=https://yum.repos.intel.com/${XPU_REPO_NAME}
 enabled=1
 gpgcheck=1
 repo_gpgcheck=1
@@ -118,7 +118,7 @@ function install_sles() {
         https://repositories.intel.com/gpu/sles/${VERSION_SP}${XPU_DRIVER_VERSION}/unified/intel-gpu-${VERSION_SP}.repo
     rpm --import https://repositories.intel.com/gpu/intel-graphics.key
     # To add the online network network package repository for the Intel Support Packages
-    zypper addrepo https://yum.repos.intel.com/oneapi oneAPI
+    zypper addrepo https://yum.repos.intel.com/${XPU_REPO_NAME} oneAPI
     rpm --import https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB
 
     # The xpu-smi packages
@@ -141,10 +141,10 @@ if [[ "${XPU_DRIVER_TYPE,,}" == "rolling" ]]; then
     XPU_DRIVER_VERSION=""
 fi
 
-# Default use Intel® oneAPI Deep Learning Essentials 2025.0
-if [[ "$XPU_VERSION" == "2025.1" ]]; then
-    XPU_PACKAGES="intel-deep-learning-essentials-2025.1"
-else
+XPU_REPO_NAME="intel-for-pytorch-gpu-dev"
+XPU_PACKAGES="intel-for-pytorch-gpu-dev-0.5 intel-pti-dev-0.9"
+if [[ "$XPU_VERSION" == "2025.0" ]]; then
+    XPU_REPO_NAME="oneapi"
     XPU_PACKAGES="intel-deep-learning-essentials-2025.0"
 fi
 

.ci/docker/libtorch/Dockerfile

@@ -51,7 +51,6 @@ ADD ./common/install_cuda.sh install_cuda.sh
 ADD ./common/install_magma.sh install_magma.sh
 COPY ./common/install_nccl.sh install_nccl.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
 ENV CUDA_HOME /usr/local/cuda
 
 FROM cuda as cuda11.8

.ci/docker/linter-cuda/Dockerfile

@@ -32,8 +32,7 @@ ARG CUDA_VERSION
 COPY ./common/install_cuda.sh install_cuda.sh
 COPY ./common/install_nccl.sh install_nccl.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
-RUN bash ./install_cuda.sh ${CUDA_VERSION} && rm install_cuda.sh install_nccl.sh /ci_commit_pins/nccl-cu* install_cusparselt.sh
+RUN bash ./install_cuda.sh ${CUDA_VERSION} && rm install_cuda.sh install_nccl.sh /ci_commit_pins/nccl-cu*
 ENV DESIRED_CUDA ${CUDA_VERSION}
 ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:$PATH
 

.ci/docker/linter/Dockerfile

@@ -16,6 +16,7 @@ RUN bash ./install_user.sh && rm install_user.sh
 
 # Install conda and other packages (e.g., numpy, pytest)
 ARG PYTHON_VERSION
+ARG PIP_CMAKE
 ENV PATH /var/lib/jenkins/ci_env/bin:$PATH
 ENV VIRTUAL_ENV /var/lib/jenkins/ci_env
 COPY requirements-ci.txt /opt/requirements-ci.txt

.ci/docker/manywheel/Dockerfile

@@ -0,0 +1,202 @@
+# syntax = docker/dockerfile:experimental
+ARG ROCM_VERSION=3.7
+ARG BASE_CUDA_VERSION=11.8
+
+ARG GPU_IMAGE=centos:7
+FROM centos:7 as base
+
+ENV LC_ALL en_US.UTF-8
+ENV LANG en_US.UTF-8
+ENV LANGUAGE en_US.UTF-8
+
+ARG DEVTOOLSET_VERSION=9
+
+# Note: This is required patch since CentOS have reached EOL
+# otherwise any yum install setp will fail
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
+RUN yum install -y wget curl perl util-linux xz bzip2 git patch which perl zlib-devel
+# Just add everything as a safe.directory for git since these will be used in multiple places with git
+RUN git config --global --add safe.directory '*'
+RUN yum install -y yum-utils centos-release-scl
+RUN yum-config-manager --enable rhel-server-rhscl-7-rpms
+# Note: After running yum-config-manager --enable rhel-server-rhscl-7-rpms
+# patch is required once again. Somehow this steps adds mirror.centos.org
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
+RUN yum install -y devtoolset-${DEVTOOLSET_VERSION}-gcc devtoolset-${DEVTOOLSET_VERSION}-gcc-c++ devtoolset-${DEVTOOLSET_VERSION}-gcc-gfortran devtoolset-${DEVTOOLSET_VERSION}-binutils
+ENV PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
+ENV LD_LIBRARY_PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib:$LD_LIBRARY_PATH
+
+RUN yum --enablerepo=extras install -y epel-release
+
+# cmake-3.18.4 from pip
+RUN yum install -y python3-pip && \
+    python3 -mpip install cmake==3.18.4 && \
+    ln -s /usr/local/bin/cmake /usr/bin/cmake
+
+RUN yum install -y autoconf aclocal automake make sudo
+
+FROM base as openssl
+# Install openssl (this must precede `build python` step)
+# (In order to have a proper SSL module, Python is compiled
+# against a recent openssl [see env vars above], which is linked
+# statically. We delete openssl afterwards.)
+ADD ./common/install_openssl.sh install_openssl.sh
+RUN bash ./install_openssl.sh && rm install_openssl.sh
+
+# EPEL for cmake
+FROM base as patchelf
+# Install patchelf
+ADD ./common/install_patchelf.sh install_patchelf.sh
+RUN bash ./install_patchelf.sh && rm install_patchelf.sh
+RUN cp $(which patchelf) /patchelf
+
+FROM patchelf as python
+# build python
+COPY manywheel/build_scripts /build_scripts
+ADD ./common/install_cpython.sh /build_scripts/install_cpython.sh
+RUN bash build_scripts/build.sh && rm -r build_scripts
+
+FROM base as cuda
+ARG BASE_CUDA_VERSION=10.2
+# Install CUDA
+ADD ./common/install_cuda.sh install_cuda.sh
+COPY ./common/install_nccl.sh install_nccl.sh
+COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
+RUN bash ./install_cuda.sh ${BASE_CUDA_VERSION} && rm install_cuda.sh install_nccl.sh /ci_commit_pins/nccl-cu*
+
+FROM base as intel
+# MKL
+ADD ./common/install_mkl.sh install_mkl.sh
+RUN bash ./install_mkl.sh && rm install_mkl.sh
+
+FROM base as magma
+ARG BASE_CUDA_VERSION=10.2
+# Install magma
+ADD ./common/install_magma.sh install_magma.sh
+RUN bash ./install_magma.sh ${BASE_CUDA_VERSION} && rm install_magma.sh
+
+FROM base as jni
+# Install java jni header
+ADD ./common/install_jni.sh install_jni.sh
+ADD ./java/jni.h jni.h
+RUN bash ./install_jni.sh && rm install_jni.sh
+
+FROM base as libpng
+# Install libpng
+ADD ./common/install_libpng.sh install_libpng.sh
+RUN bash ./install_libpng.sh && rm install_libpng.sh
+
+FROM ${GPU_IMAGE} as common
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
+ENV LC_ALL en_US.UTF-8
+ENV LANG en_US.UTF-8
+ENV LANGUAGE en_US.UTF-8
+RUN yum install -y \
+        aclocal \
+        autoconf \
+        automake \
+        bison \
+        bzip2 \
+        curl \
+        diffutils \
+        file \
+        git \
+        make \
+        patch \
+        perl \
+        unzip \
+        util-linux \
+        wget \
+        which \
+        xz \
+        yasm
+RUN yum install -y \
+    https://repo.ius.io/ius-release-el7.rpm \
+    https://ossci-linux.s3.amazonaws.com/epel-release-7-14.noarch.rpm
+
+RUN yum swap -y git git236-core
+# git236+ would refuse to run git commands in repos owned by other users
+# Which causes version check to fail, as pytorch repo is bind-mounted into the image
+# Override this behaviour by treating every folder as safe
+# For more details see https://github.com/pytorch/pytorch/issues/78659#issuecomment-1144107327
+RUN git config --global --add safe.directory "*"
+
+ENV SSL_CERT_FILE=/opt/_internal/certs.pem
+# Install LLVM version
+COPY --from=openssl            /opt/openssl                          /opt/openssl
+COPY --from=python             /opt/python                           /opt/python
+COPY --from=python             /opt/_internal                        /opt/_internal
+COPY --from=python             /opt/python/cp39-cp39/bin/auditwheel /usr/local/bin/auditwheel
+COPY --from=intel              /opt/intel                            /opt/intel
+COPY --from=patchelf           /usr/local/bin/patchelf               /usr/local/bin/patchelf
+COPY --from=jni                /usr/local/include/jni.h              /usr/local/include/jni.h
+COPY --from=libpng             /usr/local/bin/png*                   /usr/local/bin/
+COPY --from=libpng             /usr/local/bin/libpng*                /usr/local/bin/
+COPY --from=libpng             /usr/local/include/png*               /usr/local/include/
+COPY --from=libpng             /usr/local/include/libpng*            /usr/local/include/
+COPY --from=libpng             /usr/local/lib/libpng*                /usr/local/lib/
+COPY --from=libpng             /usr/local/lib/pkgconfig              /usr/local/lib/pkgconfig
+
+FROM common as cpu_final
+ARG BASE_CUDA_VERSION=10.1
+ARG DEVTOOLSET_VERSION=9
+# Install Anaconda
+ADD ./common/install_conda_docker.sh install_conda.sh
+RUN bash ./install_conda.sh && rm install_conda.sh
+ENV PATH /opt/conda/bin:$PATH
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
+
+RUN yum install -y yum-utils centos-release-scl
+RUN yum-config-manager --enable rhel-server-rhscl-7-rpms
+RUN sed -i s/mirror.centos.org/vault.centos.org/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^#.*baseurl=http/baseurl=http/g /etc/yum.repos.d/*.repo
+RUN sed -i s/^mirrorlist=http/#mirrorlist=http/g /etc/yum.repos.d/*.repo
+RUN yum install -y devtoolset-${DEVTOOLSET_VERSION}-gcc devtoolset-${DEVTOOLSET_VERSION}-gcc-c++ devtoolset-${DEVTOOLSET_VERSION}-gcc-gfortran devtoolset-${DEVTOOLSET_VERSION}-binutils
+ENV PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
+ENV LD_LIBRARY_PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib:$LD_LIBRARY_PATH
+
+# cmake is already installed inside the rocm base image, so remove if present
+RUN rpm -e cmake || true
+# cmake-3.18.4 from pip
+RUN yum install -y python3-pip && \
+    python3 -mpip install cmake==3.18.4 && \
+    ln -s /usr/local/bin/cmake /usr/bin/cmake
+
+# ninja
+RUN yum install -y ninja-build
+
+FROM cpu_final as cuda_final
+RUN rm -rf /usr/local/cuda-${BASE_CUDA_VERSION}
+COPY --from=cuda     /usr/local/cuda-${BASE_CUDA_VERSION}  /usr/local/cuda-${BASE_CUDA_VERSION}
+COPY --from=magma    /usr/local/cuda-${BASE_CUDA_VERSION}  /usr/local/cuda-${BASE_CUDA_VERSION}
+RUN ln -sf /usr/local/cuda-${BASE_CUDA_VERSION} /usr/local/cuda
+ENV PATH=/usr/local/cuda/bin:$PATH
+
+FROM cpu_final as rocm_final
+ARG ROCM_VERSION=3.7
+ARG PYTORCH_ROCM_ARCH
+ENV PYTORCH_ROCM_ARCH ${PYTORCH_ROCM_ARCH}
+# Adding ROCM_PATH env var so that LoadHip.cmake (even with logic updated for ROCm6.0)
+# find HIP works for ROCm5.7. Not needed for ROCm6.0 and above.
+# Remove below when ROCm5.7 is not in support matrix anymore.
+ENV ROCM_PATH /opt/rocm
+ENV MKLROOT /opt/intel
+# No need to install ROCm as base docker image should have full ROCm install
+#ADD ./common/install_rocm.sh install_rocm.sh
+#RUN ROCM_VERSION=${ROCM_VERSION} bash ./install_rocm.sh && rm install_rocm.sh
+ADD ./common/install_rocm_drm.sh install_rocm_drm.sh
+RUN bash ./install_rocm_drm.sh && rm install_rocm_drm.sh
+# cmake3 is needed for the MIOpen build
+RUN ln -sf /usr/local/bin/cmake /usr/bin/cmake3
+ADD ./common/install_rocm_magma.sh install_rocm_magma.sh
+RUN bash ./install_rocm_magma.sh ${ROCM_VERSION} && rm install_rocm_magma.sh
+ADD ./common/install_miopen.sh install_miopen.sh
+RUN bash ./install_miopen.sh ${ROCM_VERSION} && rm install_miopen.sh

.ci/docker/manywheel/Dockerfile_2_28

@@ -7,8 +7,8 @@ ENV LC_ALL en_US.UTF-8
 ENV LANG en_US.UTF-8
 ENV LANGUAGE en_US.UTF-8
 
-ARG DEVTOOLSET_VERSION=13
-RUN yum install -y sudo wget curl perl util-linux xz bzip2 git patch which perl zlib-devel yum-utils gcc-toolset-${DEVTOOLSET_VERSION}-gcc gcc-toolset-${DEVTOOLSET_VERSION}-gcc-c++ gcc-toolset-${DEVTOOLSET_VERSION}-gcc-gfortran gcc-toolset-${DEVTOOLSET_VERSION}-gdb
+ARG DEVTOOLSET_VERSION=11
+RUN yum install -y sudo wget curl perl util-linux xz bzip2 git patch which perl zlib-devel yum-utils gcc-toolset-${DEVTOOLSET_VERSION}-toolchain
 ENV PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
 ENV LD_LIBRARY_PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/lib:$LD_LIBRARY_PATH
 
@@ -33,13 +33,12 @@ RUN rm -rf /opt/python/cp33-cp33m /opt/_internal/cpython-3.3.6
 RUN rm -rf /opt/python/cp34-cp34m /opt/_internal/cpython-3.4.6
 
 FROM base as cuda
-ARG BASE_CUDA_VERSION=12.6
+ARG BASE_CUDA_VERSION=11.8
 # Install CUDA
 ADD ./common/install_cuda.sh install_cuda.sh
 COPY ./common/install_nccl.sh install_nccl.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
-RUN bash ./install_cuda.sh ${BASE_CUDA_VERSION} && rm install_cuda.sh install_nccl.sh ci_commit_pins/nccl-cu* install_cusparselt.sh
+RUN bash ./install_cuda.sh ${BASE_CUDA_VERSION} && rm install_cuda.sh install_nccl.sh ci_commit_pins/nccl-cu*
 
 FROM base as intel
 # MKL
@@ -47,7 +46,7 @@ ADD ./common/install_mkl.sh install_mkl.sh
 RUN bash ./install_mkl.sh && rm install_mkl.sh
 
 FROM base as magma
-ARG BASE_CUDA_VERSION=12.6
+ARG BASE_CUDA_VERSION=10.2
 # Install magma
 ADD ./common/install_magma.sh install_magma.sh
 RUN bash ./install_magma.sh ${BASE_CUDA_VERSION} && rm install_magma.sh
@@ -64,7 +63,7 @@ ADD ./common/install_libpng.sh install_libpng.sh
 RUN bash ./install_libpng.sh && rm install_libpng.sh
 
 FROM ${GPU_IMAGE} as common
-ARG DEVTOOLSET_VERSION=13
+ARG DEVTOOLSET_VERSION=11
 ENV LC_ALL en_US.UTF-8
 ENV LANG en_US.UTF-8
 ENV LANGUAGE en_US.UTF-8
@@ -87,12 +86,13 @@ RUN yum install -y \
         wget \
         which \
         xz \
-        glibc-langpack-en \
-        gcc-toolset-${DEVTOOLSET_VERSION}-gcc \
-        gcc-toolset-${DEVTOOLSET_VERSION}-gcc-c++ \
-        gcc-toolset-${DEVTOOLSET_VERSION}-gcc-gfortran \
-        gcc-toolset-${DEVTOOLSET_VERSION}-gdb
+        gcc-toolset-${DEVTOOLSET_VERSION}-toolchain \
+        glibc-langpack-en
+RUN yum install -y \
+    https://repo.ius.io/ius-release-el7.rpm \
+    https://ossci-linux.s3.amazonaws.com/epel-release-7-14.noarch.rpm
 
+RUN yum swap -y git git236-core
 # git236+ would refuse to run git commands in repos owned by other users
 # Which causes version check to fail, as pytorch repo is bind-mounted into the image
 # Override this behaviour by treating every folder as safe
@@ -116,8 +116,8 @@ COPY --from=libpng             /usr/local/lib/pkgconfig              /usr/local/
 COPY --from=jni                /usr/local/include/jni.h              /usr/local/include/jni.h
 
 FROM common as cpu_final
-ARG BASE_CUDA_VERSION=12.6
-ARG DEVTOOLSET_VERSION=13
+ARG BASE_CUDA_VERSION=11.8
+ARG DEVTOOLSET_VERSION=11
 # Install Anaconda
 ADD ./common/install_conda_docker.sh install_conda.sh
 RUN bash ./install_conda.sh && rm install_conda.sh
@@ -156,11 +156,8 @@ ENV ROCM_PATH /opt/rocm
 # and avoid 3.21.0 cmake+ninja issues with ninja inserting "-Wl,--no-as-needed" in LINK_FLAGS for static linker
 RUN python3 -m pip install --upgrade pip && \
     python3 -mpip install cmake==3.28.4
-# replace the libdrm in /opt/amdgpu with custom amdgpu.ids lookup path
 ADD ./common/install_rocm_drm.sh install_rocm_drm.sh
 RUN bash ./install_rocm_drm.sh && rm install_rocm_drm.sh
-# ROCm 6.4 rocm-smi depends on system drm.h header
-RUN yum install -y libdrm-devel
 ENV MKLROOT /opt/intel
 ADD ./common/install_rocm_magma.sh install_rocm_magma.sh
 RUN bash ./install_rocm_magma.sh ${ROCM_VERSION} && rm install_rocm_magma.sh
@@ -174,6 +171,6 @@ ENV XPU_DRIVER_TYPE ROLLING
 RUN python3 -m pip install --upgrade pip && \
     python3 -mpip install cmake==3.28.4
 ADD ./common/install_xpu.sh install_xpu.sh
-ENV XPU_VERSION 2025.1
+ENV XPU_VERSION 2025.0
 RUN bash ./install_xpu.sh && rm install_xpu.sh
 RUN pushd /opt/_internal && tar -xJf static-libs-for-embedding-only.tar.xz && popd

.ci/docker/manywheel/Dockerfile_2_28_aarch64

@@ -1,6 +1,7 @@
 FROM quay.io/pypa/manylinux_2_28_aarch64 as base
 
-ARG GCCTOOLSET_VERSION=13
+# Graviton needs GCC 10 or above for the build. GCC12 is the default version in almalinux-8.
+ARG GCCTOOLSET_VERSION=11
 
 # Language variabes
 ENV LC_ALL=en_US.UTF-8
@@ -35,10 +36,7 @@ RUN yum install -y \
   yasm \
   zstd \
   sudo \
-  gcc-toolset-${GCCTOOLSET_VERSION}-gcc \
-  gcc-toolset-${GCCTOOLSET_VERSION}-gcc-c++ \
-  gcc-toolset-${GCCTOOLSET_VERSION}-gcc-gfortran \
-  gcc-toolset-${GCCTOOLSET_VERSION}-gdb
+  gcc-toolset-${GCCTOOLSET_VERSION}-toolchain
 
 # (optional) Install non-default Ninja version
 ARG NINJA_VERSION

.ci/docker/manywheel/Dockerfile_aarch64

@@ -0,0 +1,94 @@
+FROM quay.io/pypa/manylinux2014_aarch64 as base
+
+
+# Graviton needs GCC 10 for the build
+ARG DEVTOOLSET_VERSION=10
+
+# Language variabes
+ENV LC_ALL=en_US.UTF-8
+ENV LANG=en_US.UTF-8
+ENV LANGUAGE=en_US.UTF-8
+
+# Installed needed OS packages. This is to support all
+# the binary builds (torch, vision, audio, text, data)
+RUN yum -y install epel-release
+RUN yum -y update
+RUN yum install -y \
+  autoconf \
+  automake \
+  bison \
+  bzip2 \
+  curl \
+  diffutils \
+  file \
+  git \
+  make \
+  patch \
+  perl \
+  unzip \
+  util-linux \
+  wget \
+  which \
+  xz \
+  yasm \
+  less \
+  zstd \
+  libgomp \
+  sudo \
+  devtoolset-${DEVTOOLSET_VERSION}-gcc \
+  devtoolset-${DEVTOOLSET_VERSION}-gcc-c++ \
+  devtoolset-${DEVTOOLSET_VERSION}-gcc-gfortran \
+  devtoolset-${DEVTOOLSET_VERSION}-binutils
+
+# Ensure the expected devtoolset is used
+ENV PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
+ENV LD_LIBRARY_PATH=/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/opt/rh/devtoolset-${DEVTOOLSET_VERSION}/root/usr/lib:$LD_LIBRARY_PATH
+
+
+# git236+ would refuse to run git commands in repos owned by other users
+# Which causes version check to fail, as pytorch repo is bind-mounted into the image
+# Override this behaviour by treating every folder as safe
+# For more details see https://github.com/pytorch/pytorch/issues/78659#issuecomment-1144107327
+RUN git config --global --add safe.directory "*"
+
+
+###############################################################################
+# libglfortran.a hack
+#
+# libgfortran.a from quay.io/pypa/manylinux2014_aarch64 is not compiled with -fPIC.
+# This causes __stack_chk_guard@@GLIBC_2.17 on pytorch build. To solve, get
+# ubuntu's libgfortran.a which is compiled with -fPIC
+# NOTE: Need a better way to get this library as Ubuntu's package can be removed by the vender, or changed
+###############################################################################
+RUN cd ~/ \
+  && curl -L -o ~/libgfortran-10-dev.deb http://ports.ubuntu.com/ubuntu-ports/pool/universe/g/gcc-10/libgfortran-10-dev_10.5.0-4ubuntu2_arm64.deb \
+  && ar x ~/libgfortran-10-dev.deb \
+  && tar --use-compress-program=unzstd -xvf data.tar.zst -C ~/ \
+  && cp -f ~/usr/lib/gcc/aarch64-linux-gnu/10/libgfortran.a /opt/rh/devtoolset-10/root/usr/lib/gcc/aarch64-redhat-linux/10/
+
+# install cmake
+RUN yum install -y cmake3 && \
+    ln -s /usr/bin/cmake3 /usr/bin/cmake
+
+FROM base as openssl
+# Install openssl (this must precede `build python` step)
+# (In order to have a proper SSL module, Python is compiled
+# against a recent openssl [see env vars above], which is linked
+# statically. We delete openssl afterwards.)
+ADD ./common/install_openssl.sh install_openssl.sh
+RUN bash ./install_openssl.sh && rm install_openssl.sh
+ENV SSL_CERT_FILE=/opt/_internal/certs.pem
+
+FROM base as openblas
+# Install openblas
+ADD ./common/install_openblas.sh install_openblas.sh
+RUN bash ./install_openblas.sh && rm install_openblas.sh
+
+FROM openssl as final
+# remove unncessary python versions
+RUN rm -rf /opt/python/cp26-cp26m /opt/_internal/cpython-2.6.9-ucs2
+RUN rm -rf /opt/python/cp26-cp26mu /opt/_internal/cpython-2.6.9-ucs4
+RUN rm -rf /opt/python/cp33-cp33m /opt/_internal/cpython-3.3.6
+RUN rm -rf /opt/python/cp34-cp34m /opt/_internal/cpython-3.4.6
+COPY --from=openblas     /opt/OpenBLAS/  /opt/OpenBLAS/
+ENV LD_LIBRARY_PATH=/opt/OpenBLAS/lib:$LD_LIBRARY_PATH

.ci/docker/manywheel/Dockerfile_cuda_aarch64

@@ -1,7 +1,7 @@
 FROM quay.io/pypa/manylinux_2_28_aarch64 as base
 
 # Cuda ARM build needs gcc 11
-ARG DEVTOOLSET_VERSION=13
+ARG DEVTOOLSET_VERSION=11
 
 # Language variables
 ENV LC_ALL=en_US.UTF-8
@@ -34,10 +34,7 @@ RUN yum install -y \
   zstd \
   libgomp \
   sudo \
-  gcc-toolset-${DEVTOOLSET_VERSION}-gcc \
-  gcc-toolset-${DEVTOOLSET_VERSION}-gcc-c++ \
-  gcc-toolset-${DEVTOOLSET_VERSION}-gcc-gfortran \
-  gcc-toolset-${DEVTOOLSET_VERSION}-gdb
+  gcc-toolset-${DEVTOOLSET_VERSION}-toolchain
 
 # Ensure the expected devtoolset is used
 ENV PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
@@ -69,11 +66,10 @@ RUN rm -rf /opt/python/cp34-cp34m /opt/_internal/cpython-3.4.6
 FROM base as cuda
 ARG BASE_CUDA_VERSION
 # Install CUDA
-ADD ./common/install_cuda.sh install_cuda.sh
+ADD ./common/install_cuda_aarch64.sh install_cuda_aarch64.sh
 COPY ./common/install_nccl.sh install_nccl.sh
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-RUN bash ./install_cuda.sh ${BASE_CUDA_VERSION} && rm install_cuda.sh install_nccl.sh ci_commit_pins/nccl-cu* install_cusparselt.sh
+RUN bash ./install_cuda_aarch64.sh ${BASE_CUDA_VERSION} && rm install_cuda_aarch64.sh install_nccl.sh ci_commit_pins/nccl-cu*
 
 FROM base as magma
 ARG BASE_CUDA_VERSION

.ci/docker/manywheel/Dockerfile_s390x

@@ -5,9 +5,7 @@ ENV LC_ALL=C.UTF-8
 ENV LANG=C.UTF-8
 ENV LANGUAGE=C.UTF-8
 
-# there is a bugfix in gcc >= 14 for precompiled headers and s390x vectorization interaction.
-# with earlier gcc versions test/inductor/test_cpu_cpp_wrapper.py will fail.
-ARG DEVTOOLSET_VERSION=14
+ARG DEVTOOLSET_VERSION=13
 # Installed needed OS packages. This is to support all
 # the binary builds (torch, vision, audio, text, data)
 RUN yum -y install epel-release
@@ -60,8 +58,7 @@ RUN yum install -y \
   libxslt-devel \
   libxml2-devel \
   openssl-devel \
-  valgrind \
-  ninja-build
+  valgrind
 
 ENV PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/bin:$PATH
 ENV LD_LIBRARY_PATH=/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/lib64:/opt/rh/gcc-toolset-${DEVTOOLSET_VERSION}/root/usr/lib:$LD_LIBRARY_PATH
@@ -106,6 +103,9 @@ CMD ["/bin/bash"]
 # install test dependencies:
 # - grpcio requires system openssl, bundled crypto fails to build
 RUN dnf install -y \
+  protobuf-devel \
+  protobuf-c-devel \
+  protobuf-lite-devel \
   hdf5-devel \
   python3-h5py \
   git
@@ -129,9 +129,6 @@ RUN pip3 install flatbuffers && \
   git clone https://github.com/microsoft/onnxruntime && \
   cd onnxruntime && git checkout v1.21.0 && \
   git submodule update --init --recursive && \
-  ./build.sh --config Release --parallel 0 --enable_pybind \
-  --build_wheel --enable_training --enable_training_apis \
-  --enable_training_ops --skip_tests --allow_running_as_root \
-  --compile_no_warning_as_error && \
+  ./build.sh --config Release --parallel 0 --enable_pybind --build_wheel --enable_training --enable_training_apis --enable_training_ops --skip_tests --allow_running_as_root && \
   pip3 install ./build/Linux/Release/dist/onnxruntime_training-*.whl && \
   cd .. && /bin/rm -rf ./onnxruntime

.ci/docker/manywheel/build.sh

@@ -32,13 +32,19 @@ case ${image} in
     manylinux2_28-builder:cpu)
         TARGET=cpu_final
         GPU_IMAGE=amd64/almalinux:8
-        DOCKER_GPU_BUILD_ARG=" --build-arg DEVTOOLSET_VERSION=13"
+        DOCKER_GPU_BUILD_ARG=" --build-arg DEVTOOLSET_VERSION=11"
         MANY_LINUX_VERSION="2_28"
         ;;
+    manylinuxaarch64-builder:cpu-aarch64)
+        TARGET=final
+        GPU_IMAGE=arm64v8/centos:7
+        DOCKER_GPU_BUILD_ARG=" --build-arg DEVTOOLSET_VERSION=10"
+        MANY_LINUX_VERSION="aarch64"
+        ;;
     manylinux2_28_aarch64-builder:cpu-aarch64)
         TARGET=final
         GPU_IMAGE=arm64v8/almalinux:8
-        DOCKER_GPU_BUILD_ARG=" --build-arg DEVTOOLSET_VERSION=13 --build-arg NINJA_VERSION=1.12.1"
+        DOCKER_GPU_BUILD_ARG=" --build-arg DEVTOOLSET_VERSION=11 --build-arg NINJA_VERSION=1.12.1"
         MANY_LINUX_VERSION="2_28_aarch64"
         ;;
     manylinuxcxx11-abi-builder:cpu-cxx11-abi)
@@ -53,27 +59,23 @@ case ${image} in
         DOCKER_GPU_BUILD_ARG=""
         MANY_LINUX_VERSION="s390x"
         ;;
-    manylinux2_28-builder:cuda11*)
+    manylinux2_28-builder:cuda*)
         TARGET=cuda_final
         GPU_IMAGE=amd64/almalinux:8
         DOCKER_GPU_BUILD_ARG="--build-arg BASE_CUDA_VERSION=${GPU_ARCH_VERSION} --build-arg DEVTOOLSET_VERSION=11"
         MANY_LINUX_VERSION="2_28"
         ;;
-    manylinux2_28-builder:cuda12*)
-        TARGET=cuda_final
-        GPU_IMAGE=amd64/almalinux:8
-        DOCKER_GPU_BUILD_ARG="--build-arg BASE_CUDA_VERSION=${GPU_ARCH_VERSION} --build-arg DEVTOOLSET_VERSION=13"
-        MANY_LINUX_VERSION="2_28"
-        ;;
     manylinuxaarch64-builder:cuda*)
         TARGET=cuda_final
-        GPU_IMAGE=amd64/almalinux:8
-        DOCKER_GPU_BUILD_ARG="--build-arg BASE_CUDA_VERSION=${GPU_ARCH_VERSION} --build-arg DEVTOOLSET_VERSION=13"
+        GPU_IMAGE=arm64v8/centos:7
+        DOCKER_GPU_BUILD_ARG="--build-arg BASE_CUDA_VERSION=${GPU_ARCH_VERSION} --build-arg DEVTOOLSET_VERSION=11"
         MANY_LINUX_VERSION="aarch64"
         DOCKERFILE_SUFFIX="_cuda_aarch64"
         ;;
     manylinux2_28-builder:rocm*)
         TARGET=rocm_final
+        GPU_IMAGE=rocm/dev-centos-7:${GPU_ARCH_VERSION}-complete
+        DEVTOOLSET_VERSION="9"
         MANY_LINUX_VERSION="2_28"
         DEVTOOLSET_VERSION="11"
         GPU_IMAGE=rocm/dev-almalinux-8:${GPU_ARCH_VERSION}-complete

.ci/docker/manywheel/build_scripts/build.sh

@@ -97,7 +97,7 @@ find /opt/_internal -type f -print0 \
     | xargs -0 -n1 strip --strip-unneeded 2>/dev/null || true
 # We do not need the Python test suites, or indeed the precompiled .pyc and
 # .pyo files. Partially cribbed from:
-#    https://github.com/docker-library/python/blob/master/3.4/slim/Dockerfile  # @lint-ignore
+#    https://github.com/docker-library/python/blob/master/3.4/slim/Dockerfile
 find /opt/_internal \
      \( -type d -a -name test -o -name tests \) \
   -o \( -type f -a -name '*.pyc' -o -name '*.pyo' \) \

.ci/docker/manywheel/build_scripts/build_utils.sh

@@ -2,7 +2,7 @@
 # Helper utilities for build
 # Script used only in CD pipeline
 
-OPENSSL_DOWNLOAD_URL=https://www.openssl.org/source/old/1.1.1/  # @lint-ignore
+OPENSSL_DOWNLOAD_URL=https://www.openssl.org/source/old/1.1.1/
 CURL_DOWNLOAD_URL=https://curl.se/download
 
 AUTOCONF_DOWNLOAD_URL=https://ftp.gnu.org/gnu/autoconf

.ci/docker/requirements-ci.txt

@@ -41,11 +41,14 @@ fbscribelogger==0.1.7
 #Pinned versions: 0.1.6
 #test that import:
 
-flatbuffers==24.12.23
+flatbuffers==2.0 ; platform_machine != "s390x"
 #Description: cross platform serialization library
-#Pinned versions: 24.12.23
+#Pinned versions: 2.0
 #test that import:
 
+flatbuffers ; platform_machine == "s390x"
+#Description: cross platform serialization library; Newer version is required on s390x for new python version
+
 hypothesis==5.35.1
 # Pin hypothesis to avoid flakiness: https://github.com/pytorch/pytorch/issues/31136
 #Description: advanced library for generating parametrized tests
@@ -90,7 +93,7 @@ librosa>=0.6.2 ; python_version < "3.11"
 #Pinned versions:
 #test that import:
 
-mypy==1.15.0
+mypy==1.14.0
 # Pin MyPy version because new errors are likely to appear with each release
 #Description: linter
 #Pinned versions: 1.14.0
@@ -163,10 +166,10 @@ pillow==11.0.0
 #Pinned versions: 10.3.0
 #test that import:
 
-protobuf==5.29.4
-#Description:  Google's data interchange format
-#Pinned versions: 5.29.4
-#test that import: test_tensorboard.py, test/onnx/*
+protobuf==3.20.2
+#Description:  Google’s data interchange format
+#Pinned versions: 3.20.1
+#test that import: test_tensorboard.py
 
 psutil
 #Description: information on running processes and system utilization
@@ -334,12 +337,12 @@ sympy==1.13.3
 #Pinned versions:
 #test that import:
 
-onnx==1.18.0
-#Description: Required by onnx tests, and mypy and test_public_bindings.py when checking torch.onnx._internal
+onnx==1.17.0
+#Description: Required by mypy and test_public_bindings.py when checking torch.onnx._internal
 #Pinned versions:
 #test that import:
 
-onnxscript==0.2.6
+onnxscript==0.2.2
 #Description: Required by mypy and test_public_bindings.py when checking torch.onnx._internal
 #Pinned versions:
 #test that import:
@@ -376,6 +379,3 @@ dataclasses_json==0.6.7
 #Description: required for data pipeline and scripts under tools/stats
 #Pinned versions: 0.6.7
 #test that import:
-
-cmake==4.0.0
-#Description: required for building

.ci/docker/requirements-docs.txt

@@ -1,7 +1,7 @@
 sphinx==5.3.0
 #Description: This is used to generate PyTorch docs
 #Pinned versions: 5.3.0
--e git+https://github.com/pytorch/pytorch_sphinx_theme.git@pytorch_sphinx_theme2#egg=pytorch_sphinx_theme2
+-e git+https://github.com/pytorch/pytorch_sphinx_theme.git@a98ffecb792d50df495be401becbf5c414421423#egg=pytorch_sphinx_theme2
 
 # TODO: sphinxcontrib.katex 0.9.0 adds a local KaTeX server to speed up pre-rendering
 # but it doesn't seem to work and hangs around idly. The initial thought is probably
@@ -15,10 +15,6 @@ sphinxext-opengraph==0.9.1
 #Description: This is used to generate PyTorch docs
 #Pinned versions: 0.9.1
 
-sphinx_sitemap==2.6.0
-#Description: This is used to generate sitemap for PyTorch docs
-#Pinned versions: 2.6.0
-
 matplotlib==3.5.3
 #Description: This is used to generate PyTorch docs
 #Pinned versions: 3.5.3

.ci/docker/triton_version.txt

@@ -1 +1 @@
-3.3.1
+3.3.0

.ci/docker/ubuntu-cuda/Dockerfile

@@ -26,6 +26,7 @@ RUN bash ./install_docs_reqs.sh && rm install_docs_reqs.sh
 ARG ANACONDA_PYTHON_VERSION
 ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
 ENV PATH /opt/conda/envs/py_$ANACONDA_PYTHON_VERSION/bin:/opt/conda/bin:$PATH
+ARG CONDA_CMAKE
 COPY requirements-ci.txt /opt/conda/requirements-ci.txt
 COPY ./common/install_conda.sh install_conda.sh
 COPY ./common/common_utils.sh common_utils.sh
@@ -42,6 +43,13 @@ ARG CLANG_VERSION
 COPY ./common/install_clang.sh install_clang.sh
 RUN bash ./install_clang.sh && rm install_clang.sh
 
+# (optional) Install protobuf for ONNX
+ARG PROTOBUF
+COPY ./common/install_protobuf.sh install_protobuf.sh
+RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
+RUN rm install_protobuf.sh
+ENV INSTALLED_PROTOBUF ${PROTOBUF}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./
@@ -75,6 +83,12 @@ COPY ci_commit_pins/timm.txt timm.txt
 RUN if [ -n "${INDUCTOR_BENCHMARKS}" ]; then bash ./install_inductor_benchmark_deps.sh; fi
 RUN rm install_inductor_benchmark_deps.sh common_utils.sh timm.txt huggingface.txt
 
+# (optional) Install non-default CMake version
+ARG CMAKE_VERSION
+COPY ./common/install_cmake.sh install_cmake.sh
+RUN if [ -n "${CMAKE_VERSION}" ]; then bash ./install_cmake.sh; fi
+RUN rm install_cmake.sh
+
 ARG TRITON
 
 FROM base as triton-builder

.ci/docker/ubuntu-rocm/Dockerfile

@@ -27,6 +27,7 @@ RUN bash ./install_docs_reqs.sh && rm install_docs_reqs.sh
 ARG ANACONDA_PYTHON_VERSION
 ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
 ENV PATH /opt/conda/envs/py_$ANACONDA_PYTHON_VERSION/bin:/opt/conda/bin:$PATH
+ARG CONDA_CMAKE
 COPY requirements-ci.txt /opt/conda/requirements-ci.txt
 COPY ./common/install_conda.sh install_conda.sh
 COPY ./common/common_utils.sh common_utils.sh
@@ -42,6 +43,13 @@ ARG CLANG_VERSION
 COPY ./common/install_clang.sh install_clang.sh
 RUN bash ./install_clang.sh && rm install_clang.sh
 
+# (optional) Install protobuf for ONNX
+ARG PROTOBUF
+COPY ./common/install_protobuf.sh install_protobuf.sh
+RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
+RUN rm install_protobuf.sh
+ENV INSTALLED_PROTOBUF ${PROTOBUF}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./
@@ -100,6 +108,12 @@ COPY ci_commit_pins/timm.txt timm.txt
 RUN if [ -n "${INDUCTOR_BENCHMARKS}" ]; then bash ./install_inductor_benchmark_deps.sh; fi
 RUN rm install_inductor_benchmark_deps.sh common_utils.sh timm.txt huggingface.txt
 
+# (optional) Install non-default CMake version
+ARG CMAKE_VERSION
+COPY ./common/install_cmake.sh install_cmake.sh
+RUN if [ -n "${CMAKE_VERSION}" ]; then bash ./install_cmake.sh; fi
+RUN rm install_cmake.sh
+
 # (optional) Install non-default Ninja version
 ARG NINJA_VERSION
 COPY ./common/install_ninja.sh install_ninja.sh

.ci/docker/ubuntu-xpu/Dockerfile

@@ -28,6 +28,7 @@ RUN bash ./install_docs_reqs.sh && rm install_docs_reqs.sh
 
 # Install conda and other packages (e.g., numpy, pytest)
 ARG ANACONDA_PYTHON_VERSION
+ARG CONDA_CMAKE
 ARG DOCS
 ARG BUILD_ENVIRONMENT
 ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
@@ -83,6 +84,12 @@ RUN if [ -n "${VISION}" ]; then bash ./install_vision.sh; fi
 RUN rm install_vision.sh cache_vision_models.sh common_utils.sh
 ENV INSTALLED_VISION ${VISION}
 
+# (optional) Install non-default CMake version
+ARG CMAKE_VERSION
+COPY ./common/install_cmake.sh install_cmake.sh
+RUN if [ -n "${CMAKE_VERSION}" ]; then bash ./install_cmake.sh; fi
+RUN rm install_cmake.sh
+
 # (optional) Install non-default Ninja version
 ARG NINJA_VERSION
 COPY ./common/install_ninja.sh install_ninja.sh

.ci/docker/ubuntu/Dockerfile

@@ -28,6 +28,7 @@ RUN bash ./install_docs_reqs.sh && rm install_docs_reqs.sh
 
 # Install conda and other packages (e.g., numpy, pytest)
 ARG ANACONDA_PYTHON_VERSION
+ARG CONDA_CMAKE
 ARG DOCS
 ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
 ENV PATH /opt/conda/envs/py_$ANACONDA_PYTHON_VERSION/bin:/opt/conda/bin:$PATH
@@ -53,8 +54,7 @@ ARG CUDA_VERSION
 COPY ./common/install_cuda.sh install_cuda.sh
 COPY ./common/install_nccl.sh install_nccl.sh
 COPY ./ci_commit_pins/nccl-cu* /ci_commit_pins/
-COPY ./common/install_cusparselt.sh install_cusparselt.sh
-RUN bash ./install_cuda.sh ${CUDA_VERSION} && rm install_cuda.sh install_nccl.sh /ci_commit_pins/nccl-cu* install_cusparselt.sh
+RUN bash ./install_cuda.sh ${CUDA_VERSION} && rm install_cuda.sh install_nccl.sh /ci_commit_pins/nccl-cu*
 ENV DESIRED_CUDA ${CUDA_VERSION}
 ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:$PATH
 # No effect if cuda not installed
@@ -74,6 +74,13 @@ ADD ./common/install_ucc.sh install_ucc.sh
 RUN if [ -n "${UCX_COMMIT}" ] && [ -n "${UCC_COMMIT}" ]; then bash ./install_ucc.sh; fi
 RUN rm install_ucc.sh
 
+# (optional) Install protobuf for ONNX
+ARG PROTOBUF
+COPY ./common/install_protobuf.sh install_protobuf.sh
+RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
+RUN rm install_protobuf.sh
+ENV INSTALLED_PROTOBUF ${PROTOBUF}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./
@@ -81,6 +88,12 @@ RUN if [ -n "${VISION}" ]; then bash ./install_vision.sh; fi
 RUN rm install_vision.sh cache_vision_models.sh common_utils.sh
 ENV INSTALLED_VISION ${VISION}
 
+# (optional) Install non-default CMake version
+ARG CMAKE_VERSION
+COPY ./common/install_cmake.sh install_cmake.sh
+RUN if [ -n "${CMAKE_VERSION}" ]; then bash ./install_cmake.sh; fi
+RUN rm install_cmake.sh
+
 # (optional) Install non-default Ninja version
 ARG NINJA_VERSION
 COPY ./common/install_ninja.sh install_ninja.sh

.ci/magma-rocm/Makefile

@@ -12,12 +12,13 @@ DOCKER_RUN = set -eou pipefail; ${DOCKER_CMD} run --rm -i \
 	-w /builder \
 	-e PACKAGE_NAME=${PACKAGE_NAME}${DESIRED_ROCM_SHORT} \
 	-e DESIRED_ROCM=${DESIRED_ROCM} \
-	"pytorch/almalinux-builder:rocm${DESIRED_ROCM}" \
+	"pytorch/manylinux2_28-builder:rocm${DESIRED_ROCM}-main" \
 	magma-rocm/build_magma.sh
 
 .PHONY: all
 all: magma-rocm64
 all: magma-rocm63
+all: magma-rocm624
 
 .PHONY:
 clean:
@@ -33,3 +34,8 @@ magma-rocm64:
 magma-rocm63: DESIRED_ROCM := 6.3
 magma-rocm63:
 	$(DOCKER_RUN)
+
+.PHONY: magma-rocm624
+magma-rocm624: DESIRED_ROCM := 6.2.4
+magma-rocm624:
+	$(DOCKER_RUN)

.ci/magma/Makefile

@@ -12,12 +12,13 @@ DOCKER_RUN = set -eou pipefail; ${DOCKER_CMD} run --rm -i \
 	-e PACKAGE_NAME=${PACKAGE_NAME}${DESIRED_CUDA_SHORT} \
 	-e DESIRED_CUDA=${DESIRED_CUDA} \
 	-e CUDA_ARCH_LIST="${CUDA_ARCH_LIST}" \
-	"pytorch/almalinux-builder:cuda${DESIRED_CUDA}-main" \
+	"pytorch/manylinux2_28-builder:cuda${DESIRED_CUDA}-main" \
 	magma/build_magma.sh
 
 .PHONY: all
 all: magma-cuda128
 all: magma-cuda126
+all: magma-cuda124
 all: magma-cuda118
 
 .PHONY:
@@ -36,6 +37,11 @@ magma-cuda126: DESIRED_CUDA := 12.6
 magma-cuda126:
 	$(DOCKER_RUN)
 
+.PHONY: magma-cuda124
+magma-cuda124: DESIRED_CUDA := 12.4
+magma-cuda124:
+	$(DOCKER_RUN)
+
 .PHONY: magma-cuda118
 magma-cuda118: DESIRED_CUDA := 11.8
 magma-cuda118: CUDA_ARCH_LIST += -gencode arch=compute_37,code=sm_37

.ci/manywheel/build_common.sh

@@ -18,10 +18,12 @@ retry () {
     $*  || (sleep 1 && $*) || (sleep 2 && $*) || (sleep 4 && $*) || (sleep 8 && $*)
 }
 
-PLATFORM=""
+PLATFORM="manylinux2014_x86_64"
 # TODO move this into the Docker images
 OS_NAME=$(awk -F= '/^NAME/{print $2}' /etc/os-release)
-if [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
+if [[ "$OS_NAME" == *"CentOS Linux"* ]]; then
+    retry yum install -q -y zip openssl
+elif [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
     retry yum install -q -y zip openssl
     PLATFORM="manylinux_2_28_x86_64"
 elif [[ "$OS_NAME" == *"Red Hat Enterprise Linux"* ]]; then
@@ -34,9 +36,6 @@ elif [[ "$OS_NAME" == *"Ubuntu"* ]]; then
 
     retry apt-get update
     retry apt-get -y install zip openssl
-else
-    echo "Unknown OS: '$OS_NAME'"
-    exit 1
 fi
 
 # We use the package name to test the package by passing this to 'pip install'
@@ -80,6 +79,8 @@ if [[ -e /opt/openssl ]]; then
     export CMAKE_INCLUDE_PATH="/opt/openssl/include":$CMAKE_INCLUDE_PATH
 fi
 
+
+
 mkdir -p /tmp/$WHEELHOUSE_DIR
 
 export PATCHELF_BIN=/usr/local/bin/patchelf
@@ -320,8 +321,8 @@ for pkg in /$WHEELHOUSE_DIR/torch_no_python*.whl /$WHEELHOUSE_DIR/torch*linux*.w
             # ROCm workaround for roctracer dlopens
             if [[ "$DESIRED_CUDA" == *"rocm"* ]]; then
                 patchedpath=$(fname_without_so_number $destpath)
-            # Keep the so number for XPU dependencies and libgomp.so.1 to avoid twice load
-            elif [[ "$DESIRED_CUDA" == *"xpu"* || "$filename" == "libgomp.so.1" ]]; then
+            # Keep the so number for XPU dependencies
+            elif [[ "$DESIRED_CUDA" == *"xpu"* ]]; then
                 patchedpath=$destpath
             else
                 patchedpath=$(fname_with_sha256 $destpath)

.ci/manywheel/build_cuda.sh

@@ -15,9 +15,6 @@ export INSTALL_TEST=0 # dont install test binaries into site-packages
 export USE_CUPTI_SO=0
 export USE_CUSPARSELT=${USE_CUSPARSELT:-1} # Enable if not disabled by libtorch build
 export USE_CUFILE=${USE_CUFILE:-1}
-export USE_SYSTEM_NCCL=1
-export NCCL_INCLUDE_DIR="/usr/local/cuda/include/"
-export NCCL_LIB_DIR="/usr/local/cuda/lib64/"
 
 # Keep an array of cmake variables to add to
 if [[ -z "$CMAKE_ARGS" ]]; then
@@ -39,8 +36,10 @@ if [[ -n "$DESIRED_CUDA" ]]; then
     if [[ ${DESIRED_CUDA} =~ ^[0-9]+\.[0-9]+$ ]]; then
         CUDA_VERSION=${DESIRED_CUDA}
     else
-        # cu126, cu128 etc...
-        if [[ ${#DESIRED_CUDA} -eq 5 ]]; then
+        # cu90, cu92, cu100, cu101
+        if [[ ${#DESIRED_CUDA} -eq 4 ]]; then
+            CUDA_VERSION="${DESIRED_CUDA:2:1}.${DESIRED_CUDA:3:1}"
+        elif [[ ${#DESIRED_CUDA} -eq 5 ]]; then
             CUDA_VERSION="${DESIRED_CUDA:2:2}.${DESIRED_CUDA:4:1}"
         fi
     fi
@@ -62,6 +61,10 @@ case ${CUDA_VERSION} in
         TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST};9.0"
         EXTRA_CAFFE2_CMAKE_FLAGS+=("-DATEN_NO_TEST=ON")
         ;;
+    12.4)
+        TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST};9.0"
+        EXTRA_CAFFE2_CMAKE_FLAGS+=("-DATEN_NO_TEST=ON")
+        ;;
     11.8)
         TORCH_CUDA_ARCH_LIST="${TORCH_CUDA_ARCH_LIST};3.7;9.0"
         EXTRA_CAFFE2_CMAKE_FLAGS+=("-DATEN_NO_TEST=ON")
@@ -88,15 +91,14 @@ fi
 mkdir -p "$PYTORCH_FINAL_PACKAGE_DIR" || true
 
 OS_NAME=$(awk -F= '/^NAME/{print $2}' /etc/os-release)
-if [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
+if [[ "$OS_NAME" == *"CentOS Linux"* ]]; then
+    LIBGOMP_PATH="/usr/lib64/libgomp.so.1"
+elif [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
     LIBGOMP_PATH="/usr/lib64/libgomp.so.1"
 elif [[ "$OS_NAME" == *"Red Hat Enterprise Linux"* ]]; then
     LIBGOMP_PATH="/usr/lib64/libgomp.so.1"
 elif [[ "$OS_NAME" == *"Ubuntu"* ]]; then
     LIBGOMP_PATH="/usr/lib/x86_64-linux-gnu/libgomp.so.1"
-else
-    echo "Unknown OS: '$OS_NAME'"
-    exit 1
 fi
 
 DEPS_LIST=(
@@ -106,8 +108,26 @@ DEPS_SONAME=(
     "libgomp.so.1"
 )
 
+# CUDA 11.8 have to ship the libcusparseLt.so.0 with the binary
+# since nvidia-cusparselt-cu11 is not available in PYPI
+if [[ $USE_CUSPARSELT == "1" && $CUDA_VERSION == "11.8" ]]; then
+        DEPS_SONAME+=(
+            "libcusparseLt.so.0"
+        )
+        DEPS_LIST+=(
+            "/usr/local/cuda/lib64/libcusparseLt.so.0"
+        )
+fi
 
-# CUDA_VERSION 12.6, 12.8
+
+# Turn USE_CUFILE off for CUDA 11.8, 12.4 since nvidia-cufile-cu11 and 1.9.0.20 are
+# not available in PYPI
+if [[ $CUDA_VERSION == "11.8" || $CUDA_VERSION == "12.4" ]]; then
+    export USE_CUFILE=0
+fi
+
+
+# CUDA_VERSION 12.4, 12.6, 12.8
 if [[ $CUDA_VERSION == 12* ]]; then
     export USE_STATIC_CUDNN=0
     # Try parallelizing nvcc as well
@@ -131,8 +151,6 @@ if [[ $CUDA_VERSION == 12* ]]; then
             "/usr/local/cuda/lib64/libnvToolsExt.so.1"
             "/usr/local/cuda/lib64/libnvrtc.so.12"
             "/usr/local/cuda/lib64/libnvrtc-builtins.so"
-            "/usr/local/cuda/lib64/libcufile.so.0"
-            "/usr/local/cuda/lib64/libcufile_rdma.so.1"
         )
         DEPS_SONAME+=(
             "libcudnn_adv.so.9"
@@ -150,9 +168,17 @@ if [[ $CUDA_VERSION == 12* ]]; then
             "libnvToolsExt.so.1"
             "libnvrtc.so.12"
             "libnvrtc-builtins.so"
-            "libcufile.so.0"
-            "libcufile_rdma.so.1"
         )
+        if [[ $USE_CUFILE == 1 ]]; then
+            DEPS_LIST+=(
+                "/usr/local/cuda/lib64/libcufile.so.0"
+                "/usr/local/cuda/lib64/libcufile_rdma.so.1"
+            )
+            DEPS_SONAME+=(
+                "libcufile.so.0"
+                "libcufile_rdma.so.1"
+            )
+        fi
     else
         echo "Using nvidia libs from pypi."
         CUDA_RPATHS=(
@@ -168,38 +194,31 @@ if [[ $CUDA_VERSION == 12* ]]; then
             '$ORIGIN/../../cusparselt/lib'
             '$ORIGIN/../../nvidia/nccl/lib'
             '$ORIGIN/../../nvidia/nvtx/lib'
-            '$ORIGIN/../../nvidia/cufile/lib'
         )
+        if [[ $USE_CUFILE == 1 ]]; then
+            CUDA_RPATHS+=(
+                '$ORIGIN/../../nvidia/cufile/lib'
+            )
+        fi
         CUDA_RPATHS=$(IFS=: ; echo "${CUDA_RPATHS[*]}")
         export C_SO_RPATH=$CUDA_RPATHS':$ORIGIN:$ORIGIN/lib'
         export LIB_SO_RPATH=$CUDA_RPATHS':$ORIGIN'
         export FORCE_RPATH="--force-rpath"
         export USE_STATIC_NCCL=0
+        export USE_SYSTEM_NCCL=1
         export ATEN_STATIC_CUDA=0
         export USE_CUDA_STATIC_LINK=0
         export USE_CUPTI_SO=1
+        export NCCL_INCLUDE_DIR="/usr/local/cuda/include/"
+        export NCCL_LIB_DIR="/usr/local/cuda/lib64/"
     fi
 elif [[ $CUDA_VERSION == "11.8" ]]; then
     export USE_STATIC_CUDNN=0
-    # Turn USE_CUFILE off for CUDA 11.8 since nvidia-cufile-cu11 and 1.9.0.20 are
-    # not available in PYPI
-    export USE_CUFILE=0
     # Try parallelizing nvcc as well
     export TORCH_NVCC_FLAGS="-Xfatbin -compress-all --threads 2"
     # Bundle ptxas into the wheel, see https://github.com/pytorch/pytorch/pull/119750
     export BUILD_BUNDLE_PTXAS=1
 
-    # CUDA 11.8 have to ship the libcusparseLt.so.0 with the binary
-    # since nvidia-cusparselt-cu11 is not available in PYPI
-    if [[ $USE_CUSPARSELT == "1" ]]; then
-        DEPS_SONAME+=(
-            "libcusparseLt.so.0"
-        )
-        DEPS_LIST+=(
-            "/usr/local/cuda/lib64/libcusparseLt.so.0"
-        )
-    fi
-
     if [[ -z "$PYTORCH_EXTRA_INSTALL_REQUIREMENTS" ]]; then
         echo "Bundling with cudnn and cublas."
         DEPS_LIST+=(
@@ -254,9 +273,12 @@ elif [[ $CUDA_VERSION == "11.8" ]]; then
         export LIB_SO_RPATH=$CUDA_RPATHS':$ORIGIN'
         export FORCE_RPATH="--force-rpath"
         export USE_STATIC_NCCL=0
+        export USE_SYSTEM_NCCL=1
         export ATEN_STATIC_CUDA=0
         export USE_CUDA_STATIC_LINK=0
         export USE_CUPTI_SO=1
+        export NCCL_INCLUDE_DIR="/usr/local/cuda/include/"
+        export NCCL_LIB_DIR="/usr/local/cuda/lib64/"
     fi
 else
     echo "Unknown cuda version $CUDA_VERSION"

.ci/manywheel/build_libtorch.sh

@@ -22,7 +22,9 @@ retry () {
 
 # TODO move this into the Docker images
 OS_NAME=`awk -F= '/^NAME/{print $2}' /etc/os-release`
-if [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
+if [[ "$OS_NAME" == *"CentOS Linux"* ]]; then
+    retry yum install -q -y zip openssl
+elif [[ "$OS_NAME" == *"AlmaLinux"* ]]; then
     retry yum install -q -y zip openssl
 elif [[ "$OS_NAME" == *"Red Hat Enterprise Linux"* ]]; then
     retry dnf install -q -y zip openssl
@@ -33,9 +35,6 @@ elif [[ "$OS_NAME" == *"Ubuntu"* ]]; then
     sed -i 's/.*nvidia.*/# &/' $(find /etc/apt/ -type f -name "*.list")
     retry apt-get update
     retry apt-get -y install zip openssl
-else
-    echo "Unknown OS: '$OS_NAME'"
-    exit 1
 fi
 
 # Version: setup.py uses $PYTORCH_BUILD_VERSION.post$PYTORCH_BUILD_NUMBER if

.ci/manywheel/build_xpu.sh

@@ -20,11 +20,7 @@ fi
 source /opt/intel/oneapi/compiler/latest/env/vars.sh
 source /opt/intel/oneapi/pti/latest/env/vars.sh
 source /opt/intel/oneapi/umf/latest/env/vars.sh
-source /opt/intel/oneapi/ccl/latest/env/vars.sh
-source /opt/intel/oneapi/mpi/latest/env/vars.sh
 export USE_STATIC_MKL=1
-export USE_ONEMKL=1
-export USE_XCCL=1
 
 WHEELHOUSE_DIR="wheelhousexpu"
 LIBTORCH_HOUSE_DIR="libtorch_housexpu"

.ci/onnx/README.md

@@ -10,3 +10,5 @@ example: `py2-cuda9.0-cudnn7-ubuntu16.04`. The Docker images that are
 built on Jenkins and are used in triggered builds already have this
 environment variable set in their manifest. Also see
 `./docker/jenkins/*/Dockerfile` and search for `BUILD_ENVIRONMENT`.
+
+Our Jenkins installation is located at https://ci.pytorch.org/jenkins/.

.ci/pytorch/build.sh

@@ -171,12 +171,6 @@ fi
 if [[ "$BUILD_ENVIRONMENT" == *xpu* ]]; then
   # shellcheck disable=SC1091
   source /opt/intel/oneapi/compiler/latest/env/vars.sh
-  # shellcheck disable=SC1091
-  source /opt/intel/oneapi/ccl/latest/env/vars.sh
-  # shellcheck disable=SC1091
-  source /opt/intel/oneapi/mpi/latest/env/vars.sh
-  # Enable XCCL build
-  export USE_XCCL=1
   # XPU kineto feature dependencies are not fully ready, disable kineto build as temp WA
   export USE_KINETO=0
   export TORCH_XPU_ARCH_LIST=pvc

.ci/pytorch/check_binary.sh

@@ -302,22 +302,19 @@ except RuntimeError as e:
 fi
 
 ###############################################################################
-# Check for C++ ABI compatibility to GCC-11 - GCC 13
+# Check for C++ ABI compatibility to GCC-11
 ###############################################################################
 if [[ "$(uname)" == 'Linux' &&  "$PACKAGE_TYPE" == 'manywheel' ]]; then
   pushd /tmp
-  # Per https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Dialect-Options.html
-  # gcc-11 is ABI16, gcc-13 is ABI18, gcc-14 is ABI19
-  # gcc 11 - CUDA 11.8, xpu, rocm
-  # gcc 13 - CUDA 12.6, 12.8 and cpu
-  # Please see issue for reference: https://github.com/pytorch/pytorch/issues/152426
-  if [[ "$(uname -m)" == "s390x" ]]; then
-    cxx_abi="19"
-  elif [[ "$DESIRED_CUDA" != 'cu118' && "$DESIRED_CUDA" != 'xpu' && "$DESIRED_CUDA" != 'rocm'* ]]; then
-    cxx_abi="18"
-  else
-    cxx_abi="16"
+  # Per https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Dialect-Options.html gcc-11 is ABI16
+  # Though manylinux_2.28 should have been build with gcc-14, per
+  # https://github.com/pypa/manylinux?tab=readme-ov-file#manylinux_2_28-almalinux-8-based
+  # On s390x gcc 14 is used because it contains fix for interaction
+  # between precompiled headers and vectorization builtins.
+  # This fix is not available in earlier gcc versions.
+  # gcc-14 uses ABI19.
+  if [[ "$(uname -m)" != "s390x" ]]; then
+    python -c "import torch; exit(0 if torch._C._PYBIND11_BUILD_ABI == '_cxxabi1016' else 1)"
   fi
-  python -c "import torch; exit(0 if torch._C._PYBIND11_BUILD_ABI == '_cxxabi10${cxx_abi}' else 1)"
   popd
 fi

.ci/pytorch/common.sh

@@ -13,6 +13,10 @@ if [[ "${BUILD_ENVIRONMENT}" == *rocm* ]]; then
   # HIP_PLATFORM is auto-detected by hipcc; unset to avoid build errors
   unset HIP_PLATFORM
   export PYTORCH_TEST_WITH_ROCM=1
+  # temporary to locate some kernel issues on the CI nodes
+  export HSAKMT_DEBUG_LEVEL=4
+  # improve rccl performance for distributed tests
+  export HSA_FORCE_FINE_GRAIN_PCIE=1
 fi
 
 # TODO: Renable libtorch testing for MacOS, see https://github.com/pytorch/pytorch/issues/62598

.ci/pytorch/install_cache_xla.sh

@@ -1,50 +1,31 @@
 #!/bin/bash
 
 # Script for installing sccache on the xla build job, which uses xla's docker
-# image, which has sccache installed but doesn't write the stubs.  This is
-# mostly copied from .ci/docker/install_cache.sh.  Changes are: removing checks
-# that will always return the same thing, ex checks for for rocm, CUDA, changing
-# the path where sccache is installed, not changing /etc/environment, and not
-# installing/downloading sccache as it is already in the docker image.
+# image and doesn't have sccache installed on it.  This is mostly copied from
+# .ci/docker/install_cache.sh.  Changes are: removing checks that will always
+# return the same thing, ex checks for for rocm, CUDA, and changing the path
+# where sccache is installed, and not changing /etc/environment.
 
 set -ex -o pipefail
 
+install_binary() {
+  echo "Downloading sccache binary from S3 repo"
+  curl --retry 3 https://s3.amazonaws.com/ossci-linux/sccache -o /tmp/cache/bin/sccache
+}
+
 mkdir -p /tmp/cache/bin
+mkdir -p /tmp/cache/lib
 export PATH="/tmp/cache/bin:$PATH"
 
+install_binary
+chmod a+x /tmp/cache/bin/sccache
+
 function write_sccache_stub() {
   # Unset LD_PRELOAD for ps because of asan + ps issues
   # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90589
-  if [ "$1" == "gcc" ]; then
-    # Do not call sccache recursively when dumping preprocessor argument
-    # For some reason it's very important for the first cached nvcc invocation
-    cat >"/tmp/cache/bin/$1" <<EOF
-#!/bin/sh
-
-# sccache does not support -E flag, so we need to call the original compiler directly in order to avoid calling this wrapper recursively
-for arg in "\$@"; do
-  if [ "\$arg" = "-E" ]; then
-    exec $(which "$1") "\$@"
-  fi
-done
-
-if [ \$(env -u LD_PRELOAD ps -p \$PPID -o comm=) != sccache ]; then
-  exec sccache $(which "$1") "\$@"
-else
-  exec $(which "$1") "\$@"
-fi
-EOF
-  else
-    cat >"/tmp/cache/bin/$1" <<EOF
-#!/bin/sh
-
-if [ \$(env -u LD_PRELOAD ps -p \$PPID -o comm=) != sccache ]; then
-  exec sccache $(which "$1") "\$@"
-else
-  exec $(which "$1") "\$@"
-fi
-EOF
-  fi
+  # shellcheck disable=SC2086
+  # shellcheck disable=SC2059
+  printf "#!/bin/sh\nif [ \$(env -u LD_PRELOAD ps -p \$PPID -o comm=) != sccache ]; then\n  exec sccache $(which $1) \"\$@\"\nelse\n  exec $(which $1) \"\$@\"\nfi" > "/tmp/cache/bin/$1"
   chmod a+x "/tmp/cache/bin/$1"
 }
 

.ci/pytorch/macos-build.sh

@@ -40,7 +40,7 @@ if [[ ${BUILD_ENVIRONMENT} == *"distributed"* ]]; then
 else
   # Explicitly set USE_DISTRIBUTED=0 to align with the default build config on mac. This also serves as the sole CI config that tests
   # that building with USE_DISTRIBUTED=0 works at all. See https://github.com/pytorch/pytorch/issues/86448
-  USE_DISTRIBUTED=0 USE_OPENMP=1 MACOSX_DEPLOYMENT_TARGET=11.0 WERROR=1 BUILD_TEST=OFF USE_PYTORCH_METAL=1 python setup.py bdist_wheel --plat-name macosx_11_0_arm64
+  USE_DISTRIBUTED=0 USE_OPENMP=1 MACOSX_DEPLOYMENT_TARGET=11.0 WERROR=1 BUILD_TEST=OFF USE_PYTORCH_METAL=1 python setup.py bdist_wheel
 fi
 if which sccache > /dev/null; then
   print_sccache_stats

.ci/pytorch/macos-common.sh

@@ -20,4 +20,14 @@ print_cmake_info() {
   CONDA_INSTALLATION_DIR=$(dirname "$CMAKE_EXEC")
   # Print all libraries under cmake rpath for debugging
   ls -la "$CONDA_INSTALLATION_DIR/../lib"
+
+  export CMAKE_EXEC
+  # Explicitly add conda env lib folder to cmake rpath to address the flaky issue
+  # where cmake dependencies couldn't be found. This seems to point to how conda
+  # links $CMAKE_EXEC to its package cache when cloning a new environment
+  install_name_tool -add_rpath @executable_path/../lib "${CMAKE_EXEC}" || true
+  # Adding the rpath will invalidate cmake signature, so signing it again here
+  # to trust the executable. EXC_BAD_ACCESS (SIGKILL (Code Signature Invalid))
+  # with an exit code 137 otherwise
+  codesign -f -s - "${CMAKE_EXEC}" || true
 }

.ci/pytorch/macos-test.sh

@@ -42,16 +42,6 @@ test_python_all() {
   assert_git_not_dirty
 }
 
-test_python_mps() {
-  setup_test_python
-
-  time python test/run_test.py --verbose --mps
-  MTL_CAPTURE_ENABLED=1 ${CONDA_RUN} python3 test/test_mps.py --verbose -k test_metal_capture
-
-  assert_git_not_dirty
-}
-
-
 test_python_shard() {
   if [[ -z "$NUM_TEST_SHARDS" ]]; then
     echo "NUM_TEST_SHARDS must be defined to run a Python test shard"
@@ -165,7 +155,6 @@ test_jit_hooks() {
 torchbench_setup_macos() {
   git clone --recursive https://github.com/pytorch/vision torchvision
   git clone --recursive https://github.com/pytorch/audio torchaudio
-  brew install jpeg-turbo libpng
 
   pushd torchvision
   git fetch
@@ -180,8 +169,7 @@ torchbench_setup_macos() {
   git checkout "$(cat ../.github/ci_commit_pins/audio.txt)"
   git submodule update --init --recursive
   python setup.py clean
-  #TODO: Remove me, when figure out how to make TorchAudio find brew installed openmp
-  USE_OPENMP=0 python setup.py develop
+  python setup.py develop
   popd
 
   # Shellcheck doesn't like it when you pass no arguments to a function that can take args. See https://www.shellcheck.net/wiki/SC2120
@@ -189,8 +177,9 @@ torchbench_setup_macos() {
   checkout_install_torchbench
 }
 
-pip_benchmark_deps() {
-  python -mpip install --no-input astunparse requests cython scikit-learn
+conda_benchmark_deps() {
+  conda install -y astunparse numpy scipy ninja pyyaml setuptools cmake typing-extensions requests protobuf numba cython scikit-learn
+  conda install -y -c conda-forge librosa
 }
 
 
@@ -198,7 +187,7 @@ test_torchbench_perf() {
   print_cmake_info
 
   echo "Launching torchbench setup"
-  pip_benchmark_deps
+  conda_benchmark_deps
   torchbench_setup_macos
 
   TEST_REPORTS_DIR=$(pwd)/test/test-reports
@@ -225,7 +214,7 @@ test_torchbench_smoketest() {
   print_cmake_info
 
   echo "Launching torchbench setup"
-  pip_benchmark_deps
+  conda_benchmark_deps
   # shellcheck disable=SC2119,SC2120
   torchbench_setup_macos
 
@@ -233,8 +222,7 @@ test_torchbench_smoketest() {
   mkdir -p "$TEST_REPORTS_DIR"
 
   local device=mps
-  local models=(hf_T5 llama BERT_pytorch dcgan hf_GPT2 yolov3 resnet152 sam pytorch_unet stable_diffusion_text_encoder speech_transformer Super_SloMo doctr_det_predictor doctr_reco_predictor)
-  local hf_models=(GoogleFnet YituTechConvBert Speech2Text2ForCausalLM)
+  local models=(hf_T5 llama BERT_pytorch dcgan hf_GPT2 yolov3 resnet152 sam pytorch_unet stable_diffusion_text_encoder moco speech_transformer)
 
   for backend in eager inductor; do
 
@@ -249,21 +237,6 @@ test_torchbench_smoketest() {
         PYTHONPATH="$(pwd)"/torchbench python benchmarks/dynamo/torchbench.py \
           --performance --only "$model" --backend "$backend" --inference --devices "$device" "$dtype_arg" \
           --output "$TEST_REPORTS_DIR/inductor_${backend}_torchbench_${dtype}_inference_${device}_performance.csv" || true
-        if [ "$backend" == "inductor" ]; then
-          PYTHONPATH="$(pwd)"/torchbench python benchmarks/dynamo/torchbench.py \
-            --accuracy --only "$model" --backend "$backend" --inference --devices "$device" "$dtype_arg" \
-            --output "$TEST_REPORTS_DIR/inductor_${backend}_torchbench_${dtype}_inference_${device}_accuracy.csv" || true
-        fi
-      done
-      for model in "${hf_models[@]}"; do
-        if [ "$backend" == "inductor" ]; then
-          PYTHONPATH="$(pwd)"/torchbench python benchmarks/dynamo/huggingface.py \
-            --performance --only "$model" --backend "$backend" --inference --devices "$device" "$dtype_arg" \
-            --output "$TEST_REPORTS_DIR/inductor_${backend}_huggingface_${dtype}_inference_${device}_performance.csv" || true
-          PYTHONPATH="$(pwd)"/torchbench python benchmarks/dynamo/huggingface.py \
-            --accuracy --only "$model" --backend "$backend" --inference --devices "$device" "$dtype_arg" \
-            --output "$TEST_REPORTS_DIR/inductor_${backend}_huggingface_${dtype}_inference_${device}_accuracy.csv" || true
-        fi
       done
     done
 
@@ -290,7 +263,7 @@ test_hf_perf() {
   print_cmake_info
   TEST_REPORTS_DIR=$(pwd)/test/test-reports
   mkdir -p "$TEST_REPORTS_DIR"
-  pip_benchmark_deps
+  conda_benchmark_deps
   torchbench_setup_macos
 
   echo "Launching HuggingFace training perf run"
@@ -306,7 +279,7 @@ test_timm_perf() {
   print_cmake_info
   TEST_REPORTS_DIR=$(pwd)/test/test-reports
   mkdir -p "$TEST_REPORTS_DIR"
-  pip_benchmark_deps
+  conda_benchmark_deps
   torchbench_setup_macos
 
   echo "Launching timm training perf run"
@@ -332,8 +305,6 @@ elif [[ $TEST_CONFIG == *"perf_timm"* ]]; then
   test_timm_perf
 elif [[ $TEST_CONFIG == *"perf_smoketest"* ]]; then
   test_torchbench_smoketest
-elif [[ $TEST_CONFIG == *"mps"* ]]; then
-  test_python_mps
 elif [[ $NUM_TEST_SHARDS -gt 1 ]]; then
   test_python_shard "${SHARD_NUMBER}"
   if [[ "${SHARD_NUMBER}" == 1 ]]; then

.ci/pytorch/perf_test/common.sh

@@ -0,0 +1,22 @@
+#!/bin/bash
+set -e
+
+run_test () {
+  rm -rf test_tmp/ && mkdir test_tmp/ && cd test_tmp/
+  "$@"
+  cd .. && rm -rf test_tmp/
+}
+
+get_runtime_of_command () {
+  TIMEFORMAT=%R
+
+  # runtime=$( { time ($@ &> /dev/null); } 2>&1 1>/dev/null)
+  runtime=$( { time "$@"; } 2>&1 1>/dev/null)
+  if [[ $runtime == *"Error"* ]]; then
+    exit 1
+  fi
+  runtime=${runtime#+++ $@}
+  runtime=$(python -c "print($runtime)")
+
+  echo "$runtime"
+}

.ci/pytorch/perf_test/compare_with_baseline.py

@@ -0,0 +1,91 @@
+import argparse
+import json
+import math
+import sys
+
+
+parser = argparse.ArgumentParser()
+parser.add_argument(
+    "--test-name", dest="test_name", action="store", required=True, help="test name"
+)
+parser.add_argument(
+    "--sample-stats",
+    dest="sample_stats",
+    action="store",
+    required=True,
+    help="stats from sample",
+)
+parser.add_argument(
+    "--update",
+    action="store_true",
+    help="whether to update baseline using stats from sample",
+)
+args = parser.parse_args()
+
+test_name = args.test_name
+
+if "cpu" in test_name:
+    backend = "cpu"
+elif "gpu" in test_name:
+    backend = "gpu"
+
+data_file_path = f"../{backend}_runtime.json"
+
+with open(data_file_path) as data_file:
+    data = json.load(data_file)
+
+if test_name in data:
+    mean = float(data[test_name]["mean"])
+    sigma = float(data[test_name]["sigma"])
+else:
+    # Let the test pass if baseline number doesn't exist
+    mean = sys.maxsize
+    sigma = 0.001
+
+print("population mean: ", mean)
+print("population sigma: ", sigma)
+
+# Let the test pass if baseline number is NaN (which happened in
+# the past when we didn't have logic for catching NaN numbers)
+if math.isnan(mean) or math.isnan(sigma):
+    mean = sys.maxsize
+    sigma = 0.001
+
+sample_stats_data = json.loads(args.sample_stats)
+
+sample_mean = float(sample_stats_data["mean"])
+sample_sigma = float(sample_stats_data["sigma"])
+
+print("sample mean: ", sample_mean)
+print("sample sigma: ", sample_sigma)
+
+if math.isnan(sample_mean):
+    raise Exception("""Error: sample mean is NaN""")  # noqa: TRY002
+elif math.isnan(sample_sigma):
+    raise Exception("""Error: sample sigma is NaN""")  # noqa: TRY002
+
+z_value = (sample_mean - mean) / sigma
+
+print("z-value: ", z_value)
+
+if z_value >= 3:
+    raise Exception(  # noqa: TRY002
+        f"""\n
+z-value >= 3, there is high chance of perf regression.\n
+To reproduce this regression, run
+`cd .ci/pytorch/perf_test/ && bash {test_name}.sh` on your local machine
+and compare the runtime before/after your code change.
+"""
+    )
+else:
+    print("z-value < 3, no perf regression detected.")
+    if args.update:
+        print("We will use these numbers as new baseline.")
+        new_data_file_path = f"../new_{backend}_runtime.json"
+        with open(new_data_file_path) as new_data_file:
+            new_data = json.load(new_data_file)
+        new_data[test_name] = {}
+        new_data[test_name]["mean"] = sample_mean
+        new_data[test_name]["sigma"] = max(sample_sigma, sample_mean * 0.1)
+        with open(new_data_file_path, "w") as new_data_file:
+            json.dump(new_data, new_data_file, indent=4)

.ci/pytorch/perf_test/get_stats.py

@@ -0,0 +1,18 @@
+import json
+import sys
+
+import numpy
+
+
+sample_data_list = sys.argv[1:]
+sample_data_list = [float(v.strip()) for v in sample_data_list]
+
+sample_mean = numpy.mean(sample_data_list)
+sample_sigma = numpy.std(sample_data_list)
+
+data = {
+    "mean": sample_mean,
+    "sigma": sample_sigma,
+}
+
+print(json.dumps(data))

.ci/pytorch/perf_test/test_cpu_speed_mini_sequence_labeler.sh

@@ -0,0 +1,43 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_cpu_speed_mini_sequence_labeler () {
+  echo "Testing: mini sequence labeler, CPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/benchmark.git
+
+  cd benchmark/
+
+  git checkout 726567a455edbfda6199445922a8cfee82535664
+
+  cd scripts/mini_sequence_labeler
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python main.py)
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_cpu_speed_mini_sequence_labeler "$@"
+fi

.ci/pytorch/perf_test/test_cpu_speed_mnist.sh

@@ -0,0 +1,45 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_cpu_speed_mnist () {
+  echo "Testing: MNIST, CPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/examples.git -b perftests
+
+  cd examples/mnist
+
+  conda install -c pytorch torchvision-cpu
+
+  # Download data
+  python main.py --epochs 0
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python main.py --epochs 1 --no-log)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_cpu_speed_mnist "$@"
+fi

.ci/pytorch/perf_test/test_cpu_speed_torch.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+
+. ./common.sh
+
+test_cpu_speed_torch () {
+  echo "Testing: torch.*, CPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/yf225/perf-tests.git
+
+  if [ "$1" == "compare_with_baseline" ]; then
+    export ARGS=(--compare ../cpu_runtime.json)
+  elif [ "$1" == "compare_and_update" ]; then
+    export ARGS=(--compare ../cpu_runtime.json --update ../new_cpu_runtime.json)
+  elif [ "$1" == "update_only" ]; then
+    export ARGS=(--update ../new_cpu_runtime.json)
+  fi
+
+  if ! python perf-tests/modules/test_cpu_torch.py "${ARGS[@]}"; then
+    echo "To reproduce this regression, run \`cd .ci/pytorch/perf_test/ && bash ${FUNCNAME[0]}.sh\` on your local machine and compare the runtime before/after your code change."
+    exit 1
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_cpu_speed_torch "$@"
+fi

.ci/pytorch/perf_test/test_cpu_speed_torch_tensor.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+
+. ./common.sh
+
+test_cpu_speed_torch_tensor () {
+  echo "Testing: torch.Tensor.*, CPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/yf225/perf-tests.git
+
+  if [ "$1" == "compare_with_baseline" ]; then
+    export ARGS=(--compare ../cpu_runtime.json)
+  elif [ "$1" == "compare_and_update" ]; then
+    export ARGS=(--compare ../cpu_runtime.json --update ../new_cpu_runtime.json)
+  elif [ "$1" == "update_only" ]; then
+    export ARGS=(--update ../new_cpu_runtime.json)
+  fi
+
+  if ! python perf-tests/modules/test_cpu_torch_tensor.py "${ARGS[@]}"; then
+    echo "To reproduce this regression, run \`cd .ci/pytorch/perf_test/ && bash ${FUNCNAME[0]}.sh\` on your local machine and compare the runtime before/after your code change."
+    exit 1
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_cpu_speed_torch_tensor "$@"
+fi

.ci/pytorch/perf_test/test_gpu_speed_cudnn_lstm.sh

@@ -0,0 +1,44 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_gpu_speed_cudnn_lstm () {
+  echo "Testing: CuDNN LSTM, GPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/benchmark.git
+
+  cd benchmark/
+
+  git checkout 43dfb2c0370e70ef37f249dc09aff9f0ccd2ddb0
+
+  cd scripts/
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python cudnn_lstm.py --skip-cpu-governor-check)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_gpu_speed_cudnn_lstm "$@"
+fi

.ci/pytorch/perf_test/test_gpu_speed_lstm.sh

@@ -0,0 +1,44 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_gpu_speed_lstm () {
+  echo "Testing: LSTM, GPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/benchmark.git
+
+  cd benchmark/
+
+  git checkout 43dfb2c0370e70ef37f249dc09aff9f0ccd2ddb0
+
+  cd scripts/
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python lstm.py --skip-cpu-governor-check)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_gpu_speed_lstm "$@"
+fi

.ci/pytorch/perf_test/test_gpu_speed_mlstm.sh

@@ -0,0 +1,44 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_gpu_speed_mlstm () {
+  echo "Testing: MLSTM, GPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/benchmark.git
+
+  cd benchmark/
+
+  git checkout 43dfb2c0370e70ef37f249dc09aff9f0ccd2ddb0
+
+  cd scripts/
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python mlstm.py --skip-cpu-governor-check)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_gpu_speed_mlstm "$@"
+fi

.ci/pytorch/perf_test/test_gpu_speed_mnist.sh

@@ -0,0 +1,48 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_gpu_speed_mnist () {
+  echo "Testing: MNIST, GPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/examples.git -b perftests
+
+  cd examples/mnist
+
+  conda install -c pytorch torchvision
+
+  # Download data
+  python main.py --epochs 0
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  # Needs warm up to get accurate number
+  python main.py --epochs 1 --no-log
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python main.py --epochs 1 --no-log)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_gpu_speed_mnist "$@"
+fi

.ci/pytorch/perf_test/test_gpu_speed_word_language_model.sh

@@ -0,0 +1,53 @@
+#!/bin/bash
+set -e
+
+. ./common.sh
+
+test_gpu_speed_word_language_model () {
+  echo "Testing: word language model on Wikitext-2, GPU"
+
+  export OMP_NUM_THREADS=4
+  export MKL_NUM_THREADS=4
+
+  git clone https://github.com/pytorch/examples.git -b perftests
+
+  cd examples/word_language_model
+
+  cd data/wikitext-2
+
+  # Reduce dataset size, so that we can have more runs per test
+  sed -n '1,200p' test.txt > test_tmp.txt
+  sed -n '1,1000p' train.txt > train_tmp.txt
+  sed -n '1,200p' valid.txt > valid_tmp.txt
+
+  mv test_tmp.txt test.txt
+  mv train_tmp.txt train.txt
+  mv valid_tmp.txt valid.txt
+
+  cd ../..
+
+  SAMPLE_ARRAY=()
+  NUM_RUNS=$1
+
+  for (( i=1; i<=NUM_RUNS; i++ )) do
+    runtime=$(get_runtime_of_command python main.py --cuda --epochs 1)
+    echo "$runtime"
+    SAMPLE_ARRAY+=("${runtime}")
+  done
+
+  cd ../..
+
+  stats=$(python ../get_stats.py "${SAMPLE_ARRAY[@]}")
+  echo "Runtime stats in seconds:"
+  echo "$stats"
+
+  if [ "$2" == "compare_with_baseline" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}"
+  elif [ "$2" == "compare_and_update" ]; then
+    python ../compare_with_baseline.py --test-name "${FUNCNAME[0]}" --sample-stats "${stats}" --update
+  fi
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
+  run_test test_gpu_speed_word_language_model "$@"
+fi

.ci/pytorch/perf_test/update_commit_hash.py

@@ -0,0 +1,14 @@
+import json
+import sys
+
+
+data_file_path = sys.argv[1]
+commit_hash = sys.argv[2]
+
+with open(data_file_path) as data_file:
+    data = json.load(data_file)
+
+data["commit"] = commit_hash
+
+with open(data_file_path, "w") as data_file:
+    json.dump(data, data_file)

.ci/pytorch/run_tests.sh

@@ -76,7 +76,7 @@ fi
 # Environment initialization
 if [[ "$(uname)" == Darwin ]]; then
     # Install the testing dependencies
-    retry pip install -q future hypothesis ${NUMPY_PACKAGE} ${PROTOBUF_PACKAGE} pytest setuptools six typing_extensions pyyaml
+    retry conda install -yq future hypothesis ${NUMPY_PACKAGE} ${PROTOBUF_PACKAGE} pytest setuptools six typing_extensions pyyaml
 else
     retry pip install -qr requirements.txt || true
     retry pip install -q hypothesis protobuf pytest setuptools || true
@@ -91,6 +91,7 @@ fi
 
 echo "Testing with:"
 pip freeze
+conda list || true
 
 ##############################################################################
 # Smoke tests

.ci/pytorch/short-perf-test-cpu.sh

@@ -0,0 +1,71 @@
+#!/bin/bash
+
+SCRIPT_PARENT_DIR=$(dirname "${BASH_SOURCE[0]}")
+
+# shellcheck source=.ci/pytorch/common.sh
+source "$SCRIPT_PARENT_DIR/common.sh"
+
+cd .ci/pytorch/perf_test
+
+echo "Running CPU perf test for PyTorch..."
+
+pip install -q awscli
+
+# Set multipart_threshold to be sufficiently high, so that `aws s3 cp` is not a multipart read
+# More info at https://github.com/aws/aws-cli/issues/2321
+aws configure set default.s3.multipart_threshold 5GB
+UPSTREAM_DEFAULT_BRANCH="$(git remote show https://github.com/pytorch/pytorch.git | awk '/HEAD branch/ {print $NF}')"
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # Get current default branch commit hash
+    DEFAULT_BRANCH_COMMIT_ID=$(git log --format="%H" -n 1)
+    export DEFAULT_BRANCH_COMMIT_ID
+fi
+
+# Find the default branch commit to test against
+git remote add upstream https://github.com/pytorch/pytorch.git
+git fetch upstream
+IFS=$'\n'
+while IFS='' read -r commit_id; do
+    if aws s3 ls s3://ossci-perf-test/pytorch/cpu_runtime/"${commit_id}".json; then
+        LATEST_TESTED_COMMIT=${commit_id}
+        break
+    fi
+done < <(git rev-list upstream/"$UPSTREAM_DEFAULT_BRANCH")
+aws s3 cp s3://ossci-perf-test/pytorch/cpu_runtime/"${LATEST_TESTED_COMMIT}".json cpu_runtime.json
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # Prepare new baseline file
+    cp cpu_runtime.json new_cpu_runtime.json
+    python update_commit_hash.py new_cpu_runtime.json "${DEFAULT_BRANCH_COMMIT_ID}"
+fi
+
+# Include tests
+# shellcheck source=./perf_test/test_cpu_speed_mini_sequence_labeler.sh
+. ./test_cpu_speed_mini_sequence_labeler.sh
+# shellcheck source=./perf_test/test_cpu_speed_mnist.sh
+. ./test_cpu_speed_mnist.sh
+# shellcheck source=./perf_test/test_cpu_speed_torch.sh
+. ./test_cpu_speed_torch.sh
+# shellcheck source=./perf_test/test_cpu_speed_torch_tensor.sh
+. ./test_cpu_speed_torch_tensor.sh
+
+# Run tests
+export TEST_MODE="compare_with_baseline"
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    export TEST_MODE="compare_and_update"
+fi
+
+# Operator tests
+run_test test_cpu_speed_torch ${TEST_MODE}
+run_test test_cpu_speed_torch_tensor ${TEST_MODE}
+
+# Sample model tests
+run_test test_cpu_speed_mini_sequence_labeler 20 ${TEST_MODE}
+run_test test_cpu_speed_mnist 20 ${TEST_MODE}
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # This could cause race condition if we are testing the same default branch commit twice,
+    # but the chance of them executing this line at the same time is low.
+    aws s3 cp new_cpu_runtime.json s3://ossci-perf-test/pytorch/cpu_runtime/"${DEFAULT_BRANCH_COMMIT_ID}".json --acl public-read
+fi

.ci/pytorch/short-perf-test-gpu.sh

@@ -0,0 +1,76 @@
+#!/bin/bash
+
+# shellcheck source=./common.sh
+source "$(dirname "${BASH_SOURCE[0]}")/common.sh"
+
+pushd .ci/pytorch/perf_test
+
+echo "Running GPU perf test for PyTorch..."
+
+# Trying to uninstall PyYAML can cause problem. Workaround according to:
+# https://github.com/pypa/pip/issues/5247#issuecomment-415571153
+pip install -q awscli --ignore-installed PyYAML
+
+# Set multipart_threshold to be sufficiently high, so that `aws s3 cp` is not a multipart read
+# More info at https://github.com/aws/aws-cli/issues/2321
+aws configure set default.s3.multipart_threshold 5GB
+UPSTREAM_DEFAULT_BRANCH="$(git remote show https://github.com/pytorch/pytorch.git | awk '/HEAD branch/ {print $NF}')"
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # Get current default branch commit hash
+    DEFAULT_BRANCH_COMMIT_ID=$(git log --format="%H" -n 1)
+    export DEFAULT_BRANCH_COMMIT_ID
+fi
+
+# Find the default branch commit to test against
+git remote add upstream https://github.com/pytorch/pytorch.git
+git fetch upstream
+IFS=$'\n'
+while IFS='' read -r commit_id; do
+    if aws s3 ls s3://ossci-perf-test/pytorch/gpu_runtime/"${commit_id}".json; then
+        LATEST_TESTED_COMMIT=${commit_id}
+        break
+    fi
+done < <(git rev-list upstream/"$UPSTREAM_DEFAULT_BRANCH")
+aws s3 cp s3://ossci-perf-test/pytorch/gpu_runtime/"${LATEST_TESTED_COMMIT}".json gpu_runtime.json
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # Prepare new baseline file
+    cp gpu_runtime.json new_gpu_runtime.json
+    python update_commit_hash.py new_gpu_runtime.json "${DEFAULT_BRANCH_COMMIT_ID}"
+fi
+
+# Include tests
+# shellcheck source=./perf_test/test_gpu_speed_mnist.sh
+. ./test_gpu_speed_mnist.sh
+# shellcheck source=./perf_test/test_gpu_speed_word_language_model.sh
+. ./test_gpu_speed_word_language_model.sh
+# shellcheck source=./perf_test/test_gpu_speed_cudnn_lstm.sh
+. ./test_gpu_speed_cudnn_lstm.sh
+# shellcheck source=./perf_test/test_gpu_speed_lstm.sh
+. ./test_gpu_speed_lstm.sh
+# shellcheck source=./perf_test/test_gpu_speed_mlstm.sh
+. ./test_gpu_speed_mlstm.sh
+
+# Run tests
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    run_test test_gpu_speed_mnist 20 compare_and_update
+    run_test test_gpu_speed_word_language_model 20 compare_and_update
+    run_test test_gpu_speed_cudnn_lstm 20 compare_and_update
+    run_test test_gpu_speed_lstm 20 compare_and_update
+    run_test test_gpu_speed_mlstm 20 compare_and_update
+else
+    run_test test_gpu_speed_mnist 20 compare_with_baseline
+    run_test test_gpu_speed_word_language_model 20 compare_with_baseline
+    run_test test_gpu_speed_cudnn_lstm 20 compare_with_baseline
+    run_test test_gpu_speed_lstm 20 compare_with_baseline
+    run_test test_gpu_speed_mlstm 20 compare_with_baseline
+fi
+
+if [[ "$COMMIT_SOURCE" == "$UPSTREAM_DEFAULT_BRANCH" ]]; then
+    # This could cause race condition if we are testing the same default branch commit twice,
+    # but the chance of them executing this line at the same time is low.
+    aws s3 cp new_gpu_runtime.json s3://ossci-perf-test/pytorch/gpu_runtime/"${DEFAULT_BRANCH_COMMIT_ID}".json --acl public-read
+fi
+
+popd

.ci/pytorch/smoke_test/check_gomp.py

@@ -1,74 +0,0 @@
-import ctypes
-import os
-import sys
-from pathlib import Path
-
-
-def get_gomp_thread():
-    """
-    Retrieves the maximum number of OpenMP threads after loading the `libgomp.so.1` library
-    and the `libtorch_cpu.so` library. It then queries the
-    maximum number of threads available for OpenMP parallel regions using the
-    `omp_get_max_threads` function.
-
-    Returns:
-        int: The maximum number of OpenMP threads available.
-
-    Notes:
-        - The function assumes the default path for `libgomp.so.1` on AlmaLinux OS.
-        - The path to `libtorch_cpu.so` is constructed based on the Python executable's
-          installation directory.
-        - This function is specific to environments where PyTorch and OpenMP are used
-          together and may require adjustments for other setups.
-    """
-    python_path = Path(sys.executable).resolve()
-    python_prefix = (
-        python_path.parent.parent
-    )  # Typically goes to the Python installation root
-
-    # Get the additional ABI flags (if any); it may be an empty string.
-    abiflags = getattr(sys, "abiflags", "")
-
-    # Construct the Python directory name correctly (e.g., "python3.13t").
-    python_version = (
-        f"python{sys.version_info.major}.{sys.version_info.minor}{abiflags}"
-    )
-
-    libtorch_cpu_path = (
-        python_prefix
-        / "lib"
-        / python_version
-        / "site-packages"
-        / "torch"
-        / "lib"
-        / "libtorch_cpu.so"
-    )
-
-    # use the default gomp path of AlmaLinux OS
-    libgomp_path = "/usr/lib64/libgomp.so.1"
-
-    os.environ["GOMP_CPU_AFFINITY"] = "0-3"
-
-    libgomp = ctypes.CDLL(libgomp_path)
-    libgomp = ctypes.CDLL(libtorch_cpu_path)
-
-    libgomp.omp_get_max_threads.restype = ctypes.c_int
-    libgomp.omp_get_max_threads.argtypes = []
-
-    omp_max_threads = libgomp.omp_get_max_threads()
-    return omp_max_threads
-
-
-def main():
-    omp_max_threads = get_gomp_thread()
-    print(
-        f"omp_max_threads after loading libgomp.so and libtorch_cpu.so: {omp_max_threads}"
-    )
-    if omp_max_threads == 1:
-        raise RuntimeError(
-            "omp_max_threads is 1. Check whether libgomp.so is loaded twice."
-        )
-
-
-if __name__ == "__main__":
-    main()

.ci/pytorch/test.sh

@@ -191,10 +191,6 @@ if [[ "$BUILD_ENVIRONMENT" == *xpu* ]]; then
     # shellcheck disable=SC1091
     source /opt/intel/oneapi/umf/latest/env/vars.sh
   fi
-  # shellcheck disable=SC1091
-  source /opt/intel/oneapi/ccl/latest/env/vars.sh
-  # shellcheck disable=SC1091
-  source /opt/intel/oneapi/mpi/latest/env/vars.sh
   # Check XPU status before testing
   xpu-smi discovery
 fi
@@ -318,18 +314,6 @@ test_python() {
   assert_git_not_dirty
 }
 
-test_python_smoke() {
-  # Smoke tests for H100
-  time python test/run_test.py --include test_matmul_cuda inductor/test_fp8 inductor/test_max_autotune $PYTHON_TEST_EXTRA_OPTION --upload-artifacts-while-running
-  assert_git_not_dirty
-}
-
-test_h100_distributed() {
-  # Distributed tests at H100
-  time python test/run_test.py --include distributed/_composable/test_composability/test_pp_composability.py  $PYTHON_TEST_EXTRA_OPTION --upload-artifacts-while-running
-  assert_git_not_dirty
-}
-
 test_lazy_tensor_meta_reference_disabled() {
   export TORCH_DISABLE_FUNCTIONALIZATION_META_REFERENCE=1
   echo "Testing lazy tensor operations without meta reference"
@@ -414,15 +398,8 @@ test_inductor_aoti() {
     # We need to hipify before building again
     python3 tools/amd_build/build_amd.py
   fi
-  if [[ "$BUILD_ENVIRONMENT" == *sm86* ]]; then
-    BUILD_AOT_INDUCTOR_TEST=1 TORCH_CUDA_ARCH_LIST=8.6 USE_FLASH_ATTENTION=OFF python setup.py develop
-    # TODO: Replace me completely, as one should not use conda libstdc++, nor need special path to TORCH_LIB
-    LD_LIBRARY_PATH=/opt/conda/envs/py_3.10/lib/:${TORCH_LIB_DIR}:$LD_LIBRARY_PATH
-    CPP_TESTS_DIR="${BUILD_BIN_DIR}" python test/run_test.py --cpp --verbose -i cpp/test_aoti_abi_check cpp/test_aoti_inference -dist=loadfile
-  else
-    BUILD_AOT_INDUCTOR_TEST=1 python setup.py develop
-    CPP_TESTS_DIR="${BUILD_BIN_DIR}" LD_LIBRARY_PATH="${TORCH_LIB_DIR}" python test/run_test.py --cpp --verbose -i cpp/test_aoti_abi_check cpp/test_aoti_inference -dist=loadfile
-  fi
+  BUILD_AOT_INDUCTOR_TEST=1 python setup.py develop
+  CPP_TESTS_DIR="${BUILD_BIN_DIR}" LD_LIBRARY_PATH="${TORCH_LIB_DIR}" python test/run_test.py --cpp --verbose -i cpp/test_aoti_abi_check cpp/test_aoti_inference
 }
 
 test_inductor_cpp_wrapper_shard() {
@@ -437,11 +414,10 @@ test_inductor_cpp_wrapper_shard() {
 
   if [[ "$1" -eq "2" ]]; then
     # For now, manually put the opinfo tests in shard 2, and all other tests in
-    # shard 1.  Run all CPU tests, as well as specific GPU tests triggering past
-    # bugs, for now.
+    # shard 1.  Test specific things triggering past bugs, for now.
     python test/run_test.py \
       --include inductor/test_torchinductor_opinfo \
-      -k 'linalg or to_sparse or TestInductorOpInfoCPU' \
+      -k 'linalg or to_sparse' \
       --verbose
     exit
   fi
@@ -601,6 +577,7 @@ test_perf_for_dashboard() {
     elif [[ "${TEST_CONFIG}" == *cpu_aarch64* ]]; then
       device=cpu_aarch64
     fi
+    test_inductor_set_cpu_affinity
   elif [[ "${TEST_CONFIG}" == *cuda_a10g* ]]; then
     device=cuda_a10g
   elif [[ "${TEST_CONFIG}" == *h100* ]]; then
@@ -609,9 +586,6 @@ test_perf_for_dashboard() {
     device=rocm
   fi
 
-  # Always set CPU affinity because metrics like compilation time requires CPU
-  test_inductor_set_cpu_affinity
-
   for mode in "${modes[@]}"; do
     if [[ "$mode" == "inference" ]]; then
       dtype=bfloat16
@@ -828,7 +802,16 @@ test_inductor_torchbench_smoketest_perf() {
   done
 }
 
+test_inductor_get_core_number() {
+  if [[ "${TEST_CONFIG}" == *aarch64* ]]; then
+    echo "$(($(lscpu | grep 'Cluster(s):' | awk '{print $2}') * $(lscpu | grep 'Core(s) per cluster:' | awk '{print $4}')))"
+  else
+    echo "$(($(lscpu | grep 'Socket(s):' | awk '{print $2}') * $(lscpu | grep 'Core(s) per socket:' | awk '{print $4}')))"
+  fi
+}
+
 test_inductor_set_cpu_affinity(){
+  #set jemalloc
   JEMALLOC_LIB="$(find /usr/lib -name libjemalloc.so.2)"
   export LD_PRELOAD="$JEMALLOC_LIB":"$LD_PRELOAD"
   export MALLOC_CONF="oversize_threshold:1,background_thread:true,metadata_thp:auto,dirty_decay_ms:-1,muzzy_decay_ms:-1"
@@ -840,23 +823,14 @@ test_inductor_set_cpu_affinity(){
     export KMP_AFFINITY=granularity=fine,compact,1,0
     export KMP_BLOCKTIME=1
   fi
-
-  # Use nproc here instead of lscpu because it takes into account cgroups slice
-  cpus=$(nproc)
-  thread_per_core=$(lscpu | grep 'Thread(s) per core:' | awk '{print $4}')
-  cores=$((cpus / thread_per_core))
-
-  # Set number of cores to 16 on aarch64 for performance runs
+  cores=$(test_inductor_get_core_number)
+  # Set number of cores to 16 on Aarch64 for performance runs.
   if [[ "${TEST_CONFIG}" == *aarch64* && $cores -gt 16 ]]; then
     cores=16
   fi
   export OMP_NUM_THREADS=$cores
-
-  # Handle cgroups slice start and end CPU
-  start_cpu=$(python -c 'import os; print(min(os.sched_getaffinity(0)))')
-  # Leaving one physical CPU for other tasks
-  end_cpu=$(($(python -c 'import os; print(max(os.sched_getaffinity(0)))') - thread_per_core))
-  export TASKSET="taskset -c $start_cpu-$end_cpu"
+  end_core=$((cores-1))
+  export TASKSET="taskset -c 0-$end_core"
 }
 
 test_inductor_torchbench_cpu_smoketest_perf(){
@@ -1502,6 +1476,8 @@ test_executorch() {
   export PYTHON_EXECUTABLE=python
   export CMAKE_ARGS="-DEXECUTORCH_BUILD_PYBIND=ON -DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON"
 
+  # For llama3
+  bash examples/models/llama3_2_vision/install_requirements.sh
   # NB: We need to rebuild ExecuTorch runner here because it depends on PyTorch
   # from the PR
   bash .ci/scripts/setup-linux.sh --build-tool cmake
@@ -1544,7 +1520,7 @@ test_linux_aarch64() {
        inductor/test_inplacing_pass inductor/test_kernel_benchmark inductor/test_layout_optim \
        inductor/test_max_autotune inductor/test_memory_planning inductor/test_metrics inductor/test_multi_kernel inductor/test_pad_mm \
        inductor/test_pattern_matcher inductor/test_perf inductor/test_profiler inductor/test_select_algorithm inductor/test_smoke \
-       inductor/test_split_cat_fx_passes inductor/test_compile inductor/test_torchinductor \
+       inductor/test_split_cat_fx_passes inductor/test_standalone_compile inductor/test_torchinductor \
        inductor/test_torchinductor_codegen_dynamic_shapes inductor/test_torchinductor_dynamic_shapes inductor/test_memory \
        inductor/test_triton_cpu_backend inductor/test_triton_extension_backend inductor/test_mkldnn_pattern_matcher inductor/test_cpu_cpp_wrapper \
        --shard "$SHARD_NUMBER" "$NUM_TEST_SHARDS" --verbose
@@ -1647,7 +1623,7 @@ elif [[ "${TEST_CONFIG}" == *torchbench* ]]; then
     install_torchaudio cuda
   fi
   install_torchvision
-  TORCH_CUDA_ARCH_LIST="8.0;8.6" install_torchao
+  TORCH_CUDA_ARCH_LIST="8.0;8.6" pip_install git+https://github.com/pytorch/ao.git
   id=$((SHARD_NUMBER-1))
   # https://github.com/opencv/opencv-python/issues/885
   pip_install opencv-python==4.8.0.74
@@ -1730,10 +1706,6 @@ elif [[ "${BUILD_ENVIRONMENT}" == *xpu* ]]; then
   test_python
   test_aten
   test_xpu_bin
-elif [[ "${TEST_CONFIG}" == smoke ]]; then
-  test_python_smoke
-elif [[ "${TEST_CONFIG}" == h100_distributed ]]; then
-  test_h100_distributed
 else
   install_torchvision
   install_monkeytype

.ci/pytorch/win-test-helpers/build_pytorch.bat

@@ -37,11 +37,6 @@ call %INSTALLER_DIR%\activate_miniconda3.bat
 if errorlevel 1 goto fail
 if not errorlevel 0 goto fail
 
-:: Update CMake
-call choco upgrade -y cmake --no-progress --installargs 'ADD_CMAKE_TO_PATH=System' --apply-install-arguments-to-dependencies --version=3.27.9
-if errorlevel 1 goto fail
-if not errorlevel 0 goto fail
-
 call pip install mkl-include==2021.4.0 mkl-devel==2021.4.0
 if errorlevel 1 goto fail
 if not errorlevel 0 goto fail
@@ -93,7 +88,7 @@ set PATH=%CUDA_PATH%\bin;%CUDA_PATH%\libnvvp;%PATH%
 :cuda_build_end
 
 set DISTUTILS_USE_SDK=1
-set PATH=%TMP_DIR_WIN%\bin;C:\Program Files\CMake\bin;%PATH%
+set PATH=%TMP_DIR_WIN%\bin;%PATH%
 
 :: The latest Windows CUDA test is running on AWS G5 runner with A10G GPU
 if "%TORCH_CUDA_ARCH_LIST%" == "" set TORCH_CUDA_ARCH_LIST=8.6

.ci/pytorch/win-test-helpers/installation-helpers/install_magma.bat

@@ -24,7 +24,7 @@ if "%CUDA_SUFFIX%" == "" (
 
 if "%REBUILD%"=="" (
   if "%BUILD_ENVIRONMENT%"=="" (
-    curl --retry 3 --retry-all-errors -k https://s3.amazonaws.com/ossci-windows/magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z --output %TMP_DIR_WIN%\magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z & REM @lint-ignore
+    curl --retry 3 --retry-all-errors -k https://s3.amazonaws.com/ossci-windows/magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z --output %TMP_DIR_WIN%\magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z
   ) else (
     aws s3 cp s3://ossci-windows/magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z %TMP_DIR_WIN%\magma_2.5.4_%CUDA_SUFFIX%_%BUILD_TYPE%.7z --quiet
   )

.ci/pytorch/win-test.sh

@@ -38,7 +38,7 @@ if [[ "$BUILD_ENVIRONMENT" == *cuda* ]]; then
 fi
 
 # TODO: Move both of them to Windows AMI
-python -m pip install pytest-rerunfailures==10.3 pytest-cpp==2.3.0 tensorboard==2.13.0 protobuf==5.29.4 pytest-subtests==0.13.1
+python -m pip install pytest-rerunfailures==10.3 pytest-cpp==2.3.0 tensorboard==2.13.0 pytest-subtests==0.13.1
 
 # Install Z3 optional dependency for Windows builds.
 python -m pip install z3-solver==4.12.2.0

.ci/pytorch/windows/arm64/bootstrap_libuv.bat

@@ -7,7 +7,7 @@ if not exist "%DOWNLOADS_DIR%" mkdir %DOWNLOADS_DIR%
 if not exist "%DEPENDENCIES_DIR%" mkdir %DEPENDENCIES_DIR%
 
 :: activate visual studio
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 where cl.exe
 
 cd %DEPENDENCIES_DIR%

.ci/pytorch/windows/arm64/bootstrap_openblas.bat

@@ -7,7 +7,7 @@ if not exist "%DOWNLOADS_DIR%" mkdir %DOWNLOADS_DIR%
 if not exist "%DEPENDENCIES_DIR%" mkdir %DEPENDENCIES_DIR%
 
 :: activate visual studio
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 where cl.exe
 
 :: Clone OpenBLAS

.ci/pytorch/windows/arm64/bootstrap_tests.bat

@@ -2,7 +2,7 @@
 cd %PYTORCH_ROOT%
 
 :: activate visual studio
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 where cl.exe
 
 :: create virtual environment

.ci/pytorch/windows/arm64/build_libtorch.bat

@@ -21,7 +21,7 @@ if %ENABLE_APL% == 1 (
 )
 
 :: activate visual studio
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 where cl.exe
 
 :: change to source directory

.ci/pytorch/windows/arm64/build_pytorch.bat

@@ -21,7 +21,7 @@ if %ENABLE_APL% == 1 (
 )
 
 :: activate visual studio
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 where cl.exe
 
 :: change to source directory

.ci/pytorch/windows/arm64/smoke_test.bat

@@ -33,7 +33,7 @@ pushd tmp
 set VC_VERSION_LOWER=14
 set VC_VERSION_UPPER=36
 
-call "C:\Program Files\Microsoft Visual Studio\2022\Enterprise\VC\Auxiliary\Build\vcvarsall.bat" arm64
+call "%DEPENDENCIES_DIR%\VSBuildTools\VC\Auxiliary\Build\vcvarsall.bat" arm64
 
 set install_root=%CD%
 set INCLUDE=%INCLUDE%;%install_root%\include;%install_root%\include\torch\csrc\api\include

.ci/pytorch/windows/build_pytorch.bat

@@ -1,7 +1,7 @@
 @echo off
 
-:: This script parses args, installs required libraries (MKL, Magma, libuv)
-:: and then delegates to cpu.bat, cuda80.bat, etc.
+:: This script parses args, installs required libraries (miniconda, MKL,
+:: Magma), and then delegates to cpu.bat, cuda80.bat, etc.
 
 if not "%CUDA_VERSION%" == "" if not "%PYTORCH_BUILD_VERSION%" == "" if not "%PYTORCH_BUILD_NUMBER%" == "" goto env_end
 if "%~1"=="" goto arg_error
@@ -36,18 +36,28 @@ set DESIRED_PYTHON_PREFIX=py%DESIRED_PYTHON_PREFIX:;=;py%
 set SRC_DIR=%~dp0
 pushd %SRC_DIR%
 
+:: Install Miniconda3
+set "CONDA_HOME=%CD%\conda"
+set "tmp_conda=%CONDA_HOME%"
+set "miniconda_exe=%CD%\miniconda.exe"
+rmdir /s /q conda
+del miniconda.exe
+curl --retry 3 -k https://repo.anaconda.com/miniconda/Miniconda3-py311_23.9.0-0-Windows-x86_64.exe -o "%miniconda_exe%"
+start /wait "" "%miniconda_exe%" /S /InstallationType=JustMe /RegisterPython=0 /AddToPath=0 /D=%tmp_conda%
+if ERRORLEVEL 1 exit /b 1
 set "ORIG_PATH=%PATH%"
+set "PATH=%CONDA_HOME%;%CONDA_HOME%\scripts;%CONDA_HOME%\Library\bin;%PATH%"
 
-:: setup build environment
+:: create a new conda environment and install packages
 :try
 SET /A tries=3
 :loop
 IF %tries% LEQ 0 GOTO :exception
-call setup_build.bat
+call condaenv.bat
 IF %ERRORLEVEL% EQU 0 GOTO :done
 SET /A "tries=%tries%-1"
 :exception
-echo "Failed to setup build environment"
+echo "Failed to create conda env"
 exit /B 1
 :done
 
@@ -63,7 +73,7 @@ if "%DEBUG%" == "1" (
 if not "%CUDA_VERSION%" == "cpu" if not "%CUDA_VERSION%" == "xpu" (
     rmdir /s /q magma_%CUDA_PREFIX%_%BUILD_TYPE%
     del magma_%CUDA_PREFIX%_%BUILD_TYPE%.7z
-    curl -k https://s3.amazonaws.com/ossci-windows/magma_%MAGMA_VERSION%_%CUDA_PREFIX%_%BUILD_TYPE%.7z -o magma_%CUDA_PREFIX%_%BUILD_TYPE%.7z %= @lint-ignore =%
+    curl -k https://s3.amazonaws.com/ossci-windows/magma_%MAGMA_VERSION%_%CUDA_PREFIX%_%BUILD_TYPE%.7z -o magma_%CUDA_PREFIX%_%BUILD_TYPE%.7z
     7z x -aoa magma_%CUDA_PREFIX%_%BUILD_TYPE%.7z -omagma_%CUDA_PREFIX%_%BUILD_TYPE%
 )
 
@@ -97,20 +107,19 @@ set TH_BINARY_BUILD=1
 set INSTALL_TEST=0
 
 for %%v in (%DESIRED_PYTHON_PREFIX%) do (
-
-    :: Set Environment vars for the build
-    set "CMAKE_PREFIX_PATH=%CD%\Python\Library\;%PATH%"
-    set "PYTHON_LIB_PATH=%CD%\Python\Library\bin"
-
+    :: Activate Python Environment
+    set PYTHON_PREFIX=%%v
+    set "CONDA_LIB_PATH=%CONDA_HOME%\envs\%%v\Library\bin"
     if not "%ADDITIONAL_PATH%" == "" (
-        set "PATH=%ADDITIONAL_PATH%;%PATH%"
+        set "PATH=%ADDITIONAL_PATH%;%CONDA_HOME%\envs\%%v;%CONDA_HOME%\envs\%%v\scripts;%CONDA_HOME%\envs\%%v\Library\bin;%ORIG_PATH%"
+    ) else (
+        set "PATH=%CONDA_HOME%\envs\%%v;%CONDA_HOME%\envs\%%v\scripts;%CONDA_HOME%\envs\%%v\Library\bin;%ORIG_PATH%"
     )
-
     pip install ninja
     @setlocal
     :: Set Flags
     if not "%CUDA_VERSION%"=="cpu" if not "%CUDA_VERSION%" == "xpu" (
-        set "MAGMA_HOME=%cd%\magma_%CUDA_PREFIX%_%BUILD_TYPE%"
+        set MAGMA_HOME=%cd%\magma_%CUDA_PREFIX%_%BUILD_TYPE%
     )
     echo "Calling arch build script"
     call %CUDA_PREFIX%.bat

.ci/pytorch/windows/condaenv.bat

@@ -0,0 +1,27 @@
+IF "%DESIRED_PYTHON%"=="" (
+    echo DESIRED_PYTHON is NOT defined.
+    exit /b 1
+)
+
+:: Create a new conda environment
+setlocal EnableDelayedExpansion
+FOR %%v IN (%DESIRED_PYTHON%) DO (
+    set PYTHON_VERSION_STR=%%v
+    set PYTHON_VERSION_STR=!PYTHON_VERSION_STR:.=!
+    conda remove -n py!PYTHON_VERSION_STR! --all -y || rmdir %CONDA_HOME%\envs\py!PYTHON_VERSION_STR! /s
+    if "%%v" == "3.9" call conda create -n py!PYTHON_VERSION_STR! -y numpy=2.0.1 boto3 cmake ninja typing_extensions setuptools=72.1.0 python=%%v
+    if "%%v" == "3.10" call conda create -n py!PYTHON_VERSION_STR! -y -c=conda-forge numpy=2.0.1  boto3 cmake ninja typing_extensions setuptools=72.1.0 python=%%v
+    if "%%v" == "3.11" call conda create -n py!PYTHON_VERSION_STR! -y -c=conda-forge numpy=2.0.1  boto3 cmake ninja typing_extensions setuptools=72.1.0 python=%%v
+    if "%%v" == "3.12" call conda create -n py!PYTHON_VERSION_STR! -y -c=conda-forge numpy=2.0.1  boto3 cmake ninja typing_extensions setuptools=72.1.0 python=%%v
+    if "%%v" == "3.13" call conda create -n py!PYTHON_VERSION_STR! -y -c=conda-forge numpy=2.1.2  boto3 cmake ninja typing_extensions setuptools=72.1.0 python=%%v
+    if "%%v" == "3.13t" call conda create -n py!PYTHON_VERSION_STR! -y -c=conda-forge numpy=2.1.2 boto3 cmake ninja typing_extensions setuptools=72.1.0 python-freethreading python=3.13
+    call conda run -n py!PYTHON_VERSION_STR! pip install pyyaml
+    call conda run -n py!PYTHON_VERSION_STR! pip install mkl-include
+    call conda run -n py!PYTHON_VERSION_STR! pip install mkl-static
+)
+endlocal
+
+:: Install libuv
+conda install -y -q -c conda-forge libuv=1.39
+set libuv_ROOT=%CONDA_HOME%\Library
+echo libuv_ROOT=%libuv_ROOT%

.ci/pytorch/windows/cuda124.bat

@@ -37,7 +37,7 @@ IF "%CUDA_PATH_V124%"=="" (
 )
 
 IF "%BUILD_VISION%" == "" (
-    set TORCH_CUDA_ARCH_LIST=6.1;7.0;7.5;8.0;8.6;9.0
+    set TORCH_CUDA_ARCH_LIST=5.0;6.0;6.1;7.0;7.5;8.0;8.6;9.0
     set TORCH_NVCC_FLAGS=-Xfatbin -compress-all
 ) ELSE (
     set NVCC_FLAGS=-D__CUDA_NO_HALF_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_90,code=compute_90

.ci/pytorch/windows/cuda126.bat

@@ -37,7 +37,7 @@ IF "%CUDA_PATH_V126%"=="" (
 )
 
 IF "%BUILD_VISION%" == "" (
-    set TORCH_CUDA_ARCH_LIST=6.1;7.0;7.5;8.0;8.6;9.0
+    set TORCH_CUDA_ARCH_LIST=5.0;6.0;6.1;7.0;7.5;8.0;8.6;9.0
     set TORCH_NVCC_FLAGS=-Xfatbin -compress-all
 ) ELSE (
     set NVCC_FLAGS=-D__CUDA_NO_HALF_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_90,code=compute_90

.ci/pytorch/windows/cuda128.bat

@@ -37,7 +37,7 @@ IF "%CUDA_PATH_V128%"=="" (
 )
 
 IF "%BUILD_VISION%" == "" (
-    set TORCH_CUDA_ARCH_LIST=6.1;7.0;7.5;8.0;8.6;9.0;10.0;12.0
+    set TORCH_CUDA_ARCH_LIST=5.0;6.0;6.1;7.0;7.5;8.0;8.6;9.0;10.0;12.0
     set TORCH_NVCC_FLAGS=-Xfatbin -compress-all
 ) ELSE (
     set NVCC_FLAGS=-D__CUDA_NO_HALF_OPERATORS__ --expt-relaxed-constexpr -gencode=arch=compute_50,code=sm_50 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=compute_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_90,code=compute_90 -gencode=arch=compute_100,code=compute_100 -gencode=arch=compute_120,code=compute_120

.ci/pytorch/windows/internal/7z_install.bat

@@ -1,6 +1,6 @@
 @echo off
 
-curl -k -L "https://sourceforge.net/projects/sevenzip/files/7-Zip/18.05/7z1805-x64.exe/download" -o 7z1805-x64.exe
+curl -k https://www.7-zip.org/a/7z1805-x64.exe -O
 if errorlevel 1 exit /b 1
 
 start /wait 7z1805-x64.exe /S

.ci/pytorch/windows/internal/clone.bat

@@ -8,7 +8,7 @@ goto submodule
 
 :clone_pytorch
 
-git clone https://github.com/%PYTORCH_REPO%/%MODULE_NAME% & REM @lint-ignore
+git clone https://github.com/%PYTORCH_REPO%/%MODULE_NAME%
 
 cd %MODULE_NAME%
 

.ci/pytorch/windows/internal/copy.bat

@@ -10,7 +10,7 @@ copy "%CUDA_PATH%\bin\nvrtc*64_*.dll*" pytorch\torch\lib
 copy "%CUDA_PATH%\extras\CUPTI\lib64\cupti64_*.dll*" pytorch\torch\lib
 
 copy "C:\Program Files\NVIDIA Corporation\NvToolsExt\bin\x64\nvToolsExt64_1.dll*" pytorch\torch\lib
-copy "%PYTHON_LIB_PATH%\libiomp*5md.dll" pytorch\torch\lib
+copy "%CONDA_LIB_PATH%\libiomp*5md.dll" pytorch\torch\lib
 
 :: Should be set in build_pytorch.bat
 copy "%libuv_ROOT%\bin\uv.dll" pytorch\torch\lib

.ci/pytorch/windows/internal/copy_cpu.bat

@@ -1,3 +1,3 @@
-copy "%PYTHON_LIB_PATH%\libiomp*5md.dll" pytorch\torch\lib
+copy "%CONDA_LIB_PATH%\libiomp*5md.dll" pytorch\torch\lib
 :: Should be set in build_pytorch.bat
-copy "%libuv_ROOT%\bin\uv.dll" pytorch\torch\lib
+copy "%libuv_ROOT%\bin\uv.dll" pytorch\torch\lib

.ci/pytorch/windows/internal/cuda_install.bat

@@ -35,7 +35,7 @@ exit /b 1
 
 set CUDA_INSTALL_EXE=cuda_11.8.0_522.06_windows.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
-    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" & REM @lint-ignore
+    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     set "ARGS=cuda_profiler_api_11.8 thrust_11.8 nvcc_11.8 cuobjdump_11.8 nvprune_11.8 nvprof_11.8 cupti_11.8 cublas_11.8 cublas_dev_11.8 cudart_11.8 cufft_11.8 cufft_dev_11.8 curand_11.8 curand_dev_11.8 cusolver_11.8 cusolver_dev_11.8 cusparse_11.8 cusparse_dev_11.8 npp_11.8 npp_dev_11.8 nvrtc_11.8 nvrtc_dev_11.8 nvml_dev_11.8 nvtx_11.8"
@@ -45,7 +45,7 @@ set CUDNN_FOLDER=cudnn-windows-x86_64-9.5.0.50_cuda11-archive
 set CUDNN_LIB_FOLDER="lib"
 set "CUDNN_INSTALL_ZIP=%CUDNN_FOLDER%.zip"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
-    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" & REM @lint-ignore
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
 )
@@ -62,7 +62,7 @@ goto cuda_common
 
 set CUDA_INSTALL_EXE=cuda_12.4.0_551.61_windows.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
-    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" & REM @lint-ignore
+    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     set "ARGS=cuda_profiler_api_12.4 thrust_12.4 nvcc_12.4 cuobjdump_12.4 nvprune_12.4 nvprof_12.4 cupti_12.4 cublas_12.4 cublas_dev_12.4 cudart_12.4 cufft_12.4 cufft_dev_12.4 curand_12.4 curand_dev_12.4 cusolver_12.4 cusolver_dev_12.4 cusparse_12.4 cusparse_dev_12.4 npp_12.4 npp_dev_12.4 nvrtc_12.4 nvrtc_dev_12.4 nvml_dev_12.4 nvjitlink_12.4 nvtx_12.4"
@@ -72,7 +72,7 @@ set CUDNN_FOLDER=cudnn-windows-x86_64-9.5.0.50_cuda12-archive
 set CUDNN_LIB_FOLDER="lib"
 set "CUDNN_INSTALL_ZIP=%CUDNN_FOLDER%.zip"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
-    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" & REM @lint-ignore
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
 )
@@ -89,7 +89,7 @@ goto cuda_common
 
 set CUDA_INSTALL_EXE=cuda_12.6.2_560.94_windows.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
-    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" & REM @lint-ignore
+    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     set "ARGS=cuda_profiler_api_12.6 thrust_12.6 nvcc_12.6 cuobjdump_12.6 nvprune_12.6 nvprof_12.6 cupti_12.6 cublas_12.6 cublas_dev_12.6 cudart_12.6 cufft_12.6 cufft_dev_12.6 curand_12.6 curand_dev_12.6 cusolver_12.6 cusolver_dev_12.6 cusparse_12.6 cusparse_dev_12.6 npp_12.6 npp_dev_12.6 nvrtc_12.6 nvrtc_dev_12.6 nvml_dev_12.6 nvjitlink_12.6 nvtx_12.6"
@@ -99,7 +99,7 @@ set CUDNN_FOLDER=cudnn-windows-x86_64-9.5.0.50_cuda12-archive
 set CUDNN_LIB_FOLDER="lib"
 set "CUDNN_INSTALL_ZIP=%CUDNN_FOLDER%.zip"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
-    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" & REM @lint-ignore
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
 )
@@ -116,7 +116,7 @@ goto cuda_common
 
 set CUDA_INSTALL_EXE=cuda_12.8.0_571.96_windows.exe
 if not exist "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" (
-    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%" & REM @lint-ignore
+    curl -k -L "https://ossci-windows.s3.amazonaws.com/%CUDA_INSTALL_EXE%" --output "%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     if errorlevel 1 exit /b 1
     set "CUDA_SETUP_FILE=%SRC_DIR%\temp_build\%CUDA_INSTALL_EXE%"
     set "ARGS=cuda_profiler_api_12.8 thrust_12.8 nvcc_12.8 cuobjdump_12.8 nvprune_12.8 nvprof_12.8 cupti_12.8 cublas_12.8 cublas_dev_12.8 cudart_12.8 cufft_12.8 cufft_dev_12.8 curand_12.8 curand_dev_12.8 cusolver_12.8 cusolver_dev_12.8 cusparse_12.8 cusparse_dev_12.8 npp_12.8 npp_dev_12.8 nvrtc_12.8 nvrtc_dev_12.8 nvml_dev_12.8 nvjitlink_12.8 nvtx_12.8"
@@ -126,7 +126,7 @@ set CUDNN_FOLDER=cudnn-windows-x86_64-9.7.0.66_cuda12-archive
 set CUDNN_LIB_FOLDER="lib"
 set "CUDNN_INSTALL_ZIP=%CUDNN_FOLDER%.zip"
 if not exist "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" (
-    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%" & REM @lint-ignore
+    curl -k -L "http://s3.amazonaws.com/ossci-windows/%CUDNN_INSTALL_ZIP%" --output "%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
     if errorlevel 1 exit /b 1
     set "CUDNN_SETUP_FILE=%SRC_DIR%\temp_build\%CUDNN_INSTALL_ZIP%"
 )