set thread_work_size to 4 for unrolled kernel (#154541)

set thread_work_size to 4 for unrolled kernel (#152396)

Previous PRs enabling 8-vectorization inadvertently regressed unrolled kernel perf.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/152396
Approved by: https://github.com/BoyuanFeng, https://github.com/msaroufim, https://github.com/malfet, https://github.com/Aidyn-A, https://github.com/atalman

(cherry picked from commit adebb8b11226316d45ce114beb7ac239d0f3abdd)

Co-authored-by: Natalia Gimelshein <ngimel@meta.com>

.ci/aarch64_linux/aarch64_wheel_ci_build.py

@@ -55,9 +55,22 @@ def build_ArmComputeLibrary() -> None:
         shutil.copytree(f"{acl_checkout_dir}/{d}", f"{acl_install_dir}/{d}")
 
 
-def update_wheel(wheel_path, desired_cuda) -> None:
+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:
     """
-    Update the cuda wheel libraries
+    Package the cuda wheel libraries
     """
     folder = os.path.dirname(wheel_path)
     wheelname = os.path.basename(wheel_path)
@@ -88,30 +101,19 @@ def update_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 enable_cuda:
+
+    if "128" in desired_cuda:
         libs_to_copy += [
-            "/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",
+            "/usr/local/cuda/lib64/libnvrtc-builtins.so.12.8",
+            "/usr/local/cuda/lib64/libcufile.so.0",
+            "/usr/local/cuda/lib64/libcufile_rdma.so.1",
         ]
+
     # Copy libraries to unzipped_folder/a/lib
     for lib_path in libs_to_copy:
         lib_name = os.path.basename(lib_path)
@@ -120,6 +122,13 @@ def update_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(
@@ -136,6 +145,9 @@ def complete_wheel(folder: str) -> str:
     """
     wheel_name = list_dir(f"/{folder}/dist")[0]
 
+    # Please note for cuda we don't run auditwheel since we use custom script to package
+    # the cuda dependencies to the wheel file using update_wheel() method.
+    # However we need to make sure filename reflects the correct Manylinux platform.
     if "pytorch" in folder and not enable_cuda:
         print("Repairing Wheel with AuditWheel")
         check_call(["auditwheel", "repair", f"dist/{wheel_name}"], cwd=folder)
@@ -147,7 +159,14 @@ def complete_wheel(folder: str) -> str:
             f"/{folder}/dist/{repaired_wheel_name}",
         )
     else:
-        repaired_wheel_name = wheel_name
+        repaired_wheel_name = wheel_name.replace(
+            "linux_aarch64", "manylinux_2_28_aarch64"
+        )
+        print(f"Renaming {wheel_name} wheel to {repaired_wheel_name}")
+        os.rename(
+            f"/{folder}/dist/{wheel_name}",
+            f"/{folder}/dist/{repaired_wheel_name}",
+        )
 
     print(f"Copying {repaired_wheel_name} to artifacts")
     shutil.copy2(
@@ -232,6 +251,6 @@ if __name__ == "__main__":
         print("Updating Cuda Dependency")
         filename = os.listdir("/pytorch/dist/")
         wheel_path = f"/pytorch/dist/{filename[0]}"
-        update_wheel(wheel_path, desired_cuda)
+        package_cuda_wheel(wheel_path, desired_cuda)
     pytorch_wheel_name = complete_wheel("/pytorch/")
     print(f"Build Complete. Created {pytorch_wheel_name}..")

.ci/aarch64_linux/build_aarch64_wheel.py

@@ -19,11 +19,13 @@ import boto3
 
 # AMI images for us-east-1, change the following based on your ~/.aws/config
 os_amis = {
+    "ubuntu18_04": "ami-078eece1d8119409f",  # login_name: ubuntu
     "ubuntu20_04": "ami-052eac90edaa9d08f",  # login_name: ubuntu
     "ubuntu22_04": "ami-0c6c29c5125214c77",  # login_name: ubuntu
     "redhat8": "ami-0698b90665a2ddcf1",  # login_name: ec2-user
 }
 
+ubuntu18_04_ami = os_amis["ubuntu18_04"]
 ubuntu20_04_ami = os_amis["ubuntu20_04"]
 
 
@@ -657,6 +659,18 @@ def configure_system(
             "sudo apt-get install -y python3-dev python3-yaml python3-setuptools python3-wheel python3-pip"
         )
     host.run_cmd("pip3 install dataclasses typing-extensions")
+    # Install and switch to gcc-8 on Ubuntu-18.04
+    if not host.using_docker() and host.ami == ubuntu18_04_ami and compiler == "gcc-8":
+        host.run_cmd("sudo apt-get install -y g++-8 gfortran-8")
+        host.run_cmd(
+            "sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-8 100"
+        )
+        host.run_cmd(
+            "sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-8 100"
+        )
+        host.run_cmd(
+            "sudo update-alternatives --install /usr/bin/gfortran gfortran /usr/bin/gfortran-8 100"
+        )
     if not use_conda:
         print("Installing Cython + numpy from PyPy")
         host.run_cmd("sudo pip3 install Cython")

.ci/docker/build.sh

@@ -105,6 +105,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -118,6 +119,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -132,6 +134,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.12
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -146,6 +149,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.13
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -160,6 +164,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -173,6 +178,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -187,6 +193,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.12
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -201,6 +208,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.13
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -215,6 +223,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     UCX_COMMIT=${_UCX_COMMIT}
@@ -226,6 +235,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=10
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     CONDA_CMAKE=yes
     ONNX=yes
@@ -234,7 +244,10 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=10
     PROTOBUF=yes
+    DB=yes
     VISION=yes
+    VULKAN_SDK_VERSION=1.2.162.1
+    SWIFTSHADER=yes
     CONDA_CMAKE=yes
     TRITON=yes
     ;;
@@ -242,7 +255,10 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.11
     CLANG_VERSION=10
     PROTOBUF=yes
+    DB=yes
     VISION=yes
+    VULKAN_SDK_VERSION=1.2.162.1
+    SWIFTSHADER=yes
     CONDA_CMAKE=yes
     TRITON=yes
     ;;
@@ -250,6 +266,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=9
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     CONDA_CMAKE=yes
     TRITON=yes
@@ -258,6 +275,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     ROCM_VERSION=6.2.4
     NINJA_VERSION=1.9.0
@@ -272,6 +290,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     ROCM_VERSION=6.3
     NINJA_VERSION=1.9.0
@@ -286,6 +305,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     XPU_VERSION=0.5
     NINJA_VERSION=1.9.0
@@ -296,6 +316,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     XPU_VERSION=2025.0
     NINJA_VERSION=1.9.0
@@ -306,6 +327,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     CONDA_CMAKE=yes
@@ -319,6 +341,7 @@ case "$image" in
     CUDNN_VERSION=9
     CLANG_VERSION=12
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     TRITON=yes
     ;;
@@ -326,6 +349,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     CLANG_VERSION=12
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     CONDA_CMAKE=yes
     TRITON=yes
@@ -346,6 +370,7 @@ case "$image" in
     ANACONDA_PYTHON_VERSION=3.9
     GCC_VERSION=11
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     KATEX=yes
     CONDA_CMAKE=yes
@@ -378,19 +403,20 @@ case "$image" in
     # TODO: Use 3.9 here because of this issue https://github.com/python/mypy/issues/13627.
     # We will need to update mypy version eventually, but that's for another day. The task
     # would be to upgrade mypy to 1.0.0 with Python 3.11
-    PYTHON_VERSION=3.9
-    PIP_CMAKE=yes
+    ANACONDA_PYTHON_VERSION=3.9
+    CONDA_CMAKE=yes
     ;;
   pytorch-linux-jammy-cuda11.8-cudnn9-py3.9-linter)
-    PYTHON_VERSION=3.9
+    ANACONDA_PYTHON_VERSION=3.9
     CUDA_VERSION=11.8
-    PIP_CMAKE=yes
+    CONDA_CMAKE=yes
     ;;
   pytorch-linux-jammy-aarch64-py3.10-gcc11)
     ANACONDA_PYTHON_VERSION=3.10
     GCC_VERSION=11
     ACL=yes
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     CONDA_CMAKE=yes
     # snadampal: skipping llvm src build install because the current version
@@ -402,6 +428,7 @@ case "$image" in
     GCC_VERSION=11
     ACL=yes
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     CONDA_CMAKE=yes
     # snadampal: skipping llvm src build install because the current version
@@ -412,6 +439,7 @@ case "$image" in
   *)
     # Catch-all for builds that are not hardcoded.
     PROTOBUF=yes
+    DB=yes
     VISION=yes
     echo "image '$image' did not match an existing build configuration"
     if [[ "$image" == *py* ]]; then
@@ -467,6 +495,7 @@ docker build \
        --build-arg "BUILD_ENVIRONMENT=${image}" \
        --build-arg "PROTOBUF=${PROTOBUF:-}" \
        --build-arg "LLVMDEV=${LLVMDEV:-}" \
+       --build-arg "DB=${DB:-}" \
        --build-arg "VISION=${VISION:-}" \
        --build-arg "UBUNTU_VERSION=${UBUNTU_VERSION}" \
        --build-arg "CENTOS_VERSION=${CENTOS_VERSION}" \
@@ -474,12 +503,13 @@ docker build \
        --build-arg "GLIBC_VERSION=${GLIBC_VERSION}" \
        --build-arg "CLANG_VERSION=${CLANG_VERSION}" \
        --build-arg "ANACONDA_PYTHON_VERSION=${ANACONDA_PYTHON_VERSION}" \
-       --build-arg "PYTHON_VERSION=${PYTHON_VERSION}" \
        --build-arg "GCC_VERSION=${GCC_VERSION}" \
        --build-arg "CUDA_VERSION=${CUDA_VERSION}" \
        --build-arg "CUDNN_VERSION=${CUDNN_VERSION}" \
        --build-arg "TENSORRT_VERSION=${TENSORRT_VERSION}" \
        --build-arg "GRADLE_VERSION=${GRADLE_VERSION}" \
+       --build-arg "VULKAN_SDK_VERSION=${VULKAN_SDK_VERSION}" \
+       --build-arg "SWIFTSHADER=${SWIFTSHADER}" \
        --build-arg "CMAKE_VERSION=${CMAKE_VERSION:-}" \
        --build-arg "NINJA_VERSION=${NINJA_VERSION:-}" \
        --build-arg "KATEX=${KATEX:-}" \
@@ -489,7 +519,6 @@ docker build \
        --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}" \

.ci/docker/centos-rocm/Dockerfile

@@ -55,6 +55,13 @@ RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
 RUN rm install_protobuf.sh
 ENV INSTALLED_PROTOBUF ${PROTOBUF}
 
+# (optional) Install database packages like LMDB and LevelDB
+ARG DB
+COPY ./common/install_db.sh install_db.sh
+RUN if [ -n "${DB}" ]; then bash ./install_db.sh; fi
+RUN rm install_db.sh
+ENV INSTALLED_DB ${DB}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./

.ci/docker/ci_commit_pins/executorch.txt

@@ -1 +1 @@
-01a22b6f16d117454b7d21ebdc691b0785b84a7f
+ebe8522378c3f9944aaaef44868f5ececdd845fc

.ci/docker/common/install_base.sh

@@ -37,7 +37,7 @@ install_ubuntu() {
   if [[ "$UBUNTU_VERSION" == "20.04"* && "$CUDA_VERSION" == "11.8"* ]]; then
     maybe_libnccl_dev="libnccl2=2.15.5-1+cuda11.8 libnccl-dev=2.15.5-1+cuda11.8 --allow-downgrades --allow-change-held-packages"
   elif [[ "$UBUNTU_VERSION" == "20.04"* && "$CUDA_VERSION" == "12.4"* ]]; then
-    maybe_libnccl_dev="libnccl2=2.25.1-1+cuda12.4 libnccl-dev=2.25.1-1+cuda12.4 --allow-downgrades --allow-change-held-packages"
+    maybe_libnccl_dev="libnccl2=2.26.2-1+cuda12.4 libnccl-dev=2.26.2-1+cuda12.4 --allow-downgrades --allow-change-held-packages"
   else
     maybe_libnccl_dev=""
   fi

.ci/docker/common/install_cache.sh

@@ -2,45 +2,29 @@
 
 set -ex
 
-SCCACHE_VERSION="0.9.1"
-
-CARGO_FLAGS=""
-
-install_prereqs_ubuntu() {
+install_ubuntu() {
+  echo "Preparing to build sccache from source"
   apt-get update
   # libssl-dev will not work as it is upgraded to libssl3 in Ubuntu-22.04.
   # Instead use lib and headers from OpenSSL1.1 installed in `install_openssl.sh``
   apt-get install -y cargo
-
-  # cleanup after ourselves
-  trap 'cleanup_ubuntu' EXIT
+  echo "Checking out sccache repo"
+  git clone https://github.com/mozilla/sccache -b v0.9.1
+  cd sccache
+  echo "Building sccache"
+  cargo build --release
+  cp target/release/sccache /opt/cache/bin
+  echo "Cleaning up"
+  cd ..
+  rm -rf sccache
+  apt-get remove -y cargo rustc
+  apt-get autoclean && apt-get clean
 
   echo "Downloading old sccache binary from S3 repo for PCH builds"
   curl --retry 3 https://s3.amazonaws.com/ossci-linux/sccache -o /opt/cache/bin/sccache-0.2.14a
   chmod 755 /opt/cache/bin/sccache-0.2.14a
 }
 
-cleanup_ubuntu() {
-  rm -rf sccache
-  apt-get remove -y cargo rustc
-  apt-get autoclean && apt-get clean
-}
-
-install_prereqs_almalinux() {
-  dnf install -y cargo
-  # use vendored openssl, we're not going to use the dist-server anyways
-  CARGO_FEATURES="--bin sccache --features openssl/vendored"
-}
-
-build_and_install_sccache() {
-  # modern version of git don't like openssl1.1
-  wget -q -O sccache.tar.gz "https://github.com/mozilla/sccache/archive/refs/tags/v${SCCACHE_VERSION}.tar.gz"
-  tar xzf sccache.tar.gz
-  pushd "sccache-${SCCACHE_VERSION}"
-  cargo build --release ${CARGO_FEATURES}
-  cp target/release/sccache /opt/cache/bin
-}
-
 install_binary() {
   echo "Downloading sccache binary from S3 repo"
   curl --retry 3 https://s3.amazonaws.com/ossci-linux/sccache -o /opt/cache/bin/sccache
@@ -51,21 +35,8 @@ mkdir -p /opt/cache/lib
 sed -e 's|PATH="\(.*\)"|PATH="/opt/cache/bin:\1"|g' -i /etc/environment
 export PATH="/opt/cache/bin:$PATH"
 
-echo "Preparing to build sccache from source"
-DIST_ID=$(. /etc/os-release && echo "$ID")
-case ${DIST_ID} in
-  ubuntu)
-    install_prereqs_ubuntu
-    ;;
-  almalinux)
-    install_prereqs_almalinux
-    ;;
-  *)
-    echo "ERROR: Unknown distribution ${DIST_ID}"
-    exit 1
-    ;;
-esac
-build_and_install_sccache
+# Setup compiler cache
+install_ubuntu
 chmod a+x /opt/cache/bin/sccache
 
 function write_sccache_stub() {

.ci/docker/common/install_clang.sh

@@ -4,10 +4,16 @@ set -ex
 
 if [ -n "$CLANG_VERSION" ]; then
 
-  if [[ $UBUNTU_VERSION == 22.04 ]]; then
+  if [[ $CLANG_VERSION == 9 && $UBUNTU_VERSION == 18.04 ]]; then
+    sudo apt-get update
+    # gpg-agent is not available by default on 18.04
+    sudo apt-get install  -y --no-install-recommends gpg-agent
+    wget --no-check-certificate -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add  -
+    apt-add-repository "deb http://apt.llvm.org/bionic/ llvm-toolchain-bionic-${CLANG_VERSION} main"
+  elif [[ $UBUNTU_VERSION == 22.04 ]]; then
     # work around ubuntu apt-get conflicts
     sudo apt-get -y -f install
-    wget --no-check-certificate -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+    wget --no-check-certificate -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add  -
     if [[ $CLANG_VERSION == 18 ]]; then
       apt-add-repository "deb http://apt.llvm.org/jammy/ llvm-toolchain-jammy-18 main"
     fi
@@ -35,7 +41,7 @@ if [ -n "$CLANG_VERSION" ]; then
   # clang's packaging is a little messed up (the runtime libs aren't
   # added into the linker path), so give it a little help
   clang_lib=("/usr/lib/llvm-$CLANG_VERSION/lib/clang/"*"/lib/linux")
-  echo "$clang_lib" >/etc/ld.so.conf.d/clang.conf
+  echo "$clang_lib" > /etc/ld.so.conf.d/clang.conf
   ldconfig
 
   # Cleanup package manager

.ci/docker/common/install_cuda.sh

@@ -2,7 +2,7 @@
 
 set -ex
 
-NCCL_VERSION=v2.25.1-1
+NCCL_VERSION=v2.26.2-1
 CUDNN_VERSION=9.5.1.17
 
 function install_cusparselt_040 {
@@ -240,7 +240,7 @@ function prune_126 {
 }
 
 function install_128 {
-  CUDNN_VERSION=9.8.0.87
+  CUDNN_VERSION=9.7.1.26
   echo "Installing CUDA 12.8.0 and cuDNN ${CUDNN_VERSION} and NCCL ${NCCL_VERSION} and cuSparseLt-0.6.3"
   rm -rf /usr/local/cuda-12.8 /usr/local/cuda
   # install CUDA 12.8.0 in the same container

.ci/docker/common/install_cudnn.sh

@@ -5,7 +5,7 @@ if [[ -n "${CUDNN_VERSION}" ]]; then
     mkdir tmp_cudnn
     pushd tmp_cudnn
     if [[ ${CUDA_VERSION:0:4} == "12.8" ]]; then
-        CUDNN_NAME="cudnn-linux-x86_64-9.8.0.87_cuda12-archive"
+        CUDNN_NAME="cudnn-linux-x86_64-9.7.1.26_cuda12-archive"
     elif [[ ${CUDA_VERSION:0:4} == "12.6" ]]; then
         CUDNN_NAME="cudnn-linux-x86_64-9.5.1.17_cuda12-archive"
     elif [[ ${CUDA_VERSION:0:2} == "12" ]]; then

.ci/docker/common/install_db.sh

@@ -0,0 +1,38 @@
+#!/bin/bash
+
+set -ex
+
+install_ubuntu() {
+  apt-get update
+
+  # Cleanup
+  apt-get autoclean && apt-get clean
+  rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+}
+
+install_centos() {
+  # Need EPEL for many packages we depend on.
+  # See http://fedoraproject.org/wiki/EPEL
+  yum --enablerepo=extras install -y epel-release
+
+  # Cleanup
+  yum clean all
+  rm -rf /var/cache/yum
+  rm -rf /var/lib/yum/yumdb
+  rm -rf /var/lib/yum/history
+}
+
+# Install base packages depending on the base OS
+ID=$(grep -oP '(?<=^ID=).+' /etc/os-release | tr -d '"')
+case "$ID" in
+  ubuntu)
+    install_ubuntu
+    ;;
+  centos)
+    install_centos
+    ;;
+  *)
+    echo "Unable to determine OS..."
+    exit 1
+    ;;
+esac

.ci/docker/common/install_executorch.sh

@@ -50,8 +50,7 @@ setup_executorch() {
   pushd executorch
 
   export PYTHON_EXECUTABLE=python
-  export EXECUTORCH_BUILD_PYBIND=ON
-  export CMAKE_ARGS="-DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON"
+  export CMAKE_ARGS="-DEXECUTORCH_BUILD_PYBIND=ON -DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON"
 
   as_jenkins .ci/scripts/setup-linux.sh --build-tool cmake || true
   popd

.ci/docker/common/install_halide.sh

@@ -35,7 +35,9 @@ git clone https://github.com/halide/Halide.git
 pushd Halide
 git checkout ${COMMIT} && git submodule update --init --recursive
 pip_install -r requirements.txt
-cmake -G Ninja -DCMAKE_BUILD_TYPE=Release -S . -B build
+# NOTE: pybind has a requirement for cmake > 3.5 so set the minimum cmake version here with a flag
+#       Context: https://github.com/pytorch/pytorch/issues/150420
+cmake -G Ninja -DCMAKE_POLICY_VERSION_MINIMUM=3.5 -DCMAKE_BUILD_TYPE=Release -S . -B build
 cmake --build build
 test -e ${CONDA_PREFIX}/lib/python3 || ln -s python${ANACONDA_PYTHON_VERSION} ${CONDA_PREFIX}/lib/python3
 cmake --install build --prefix ${CONDA_PREFIX}

.ci/docker/common/install_inductor_benchmark_deps.sh

@@ -14,13 +14,6 @@ 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 cmake torch torchvision triton

.ci/docker/common/install_linter.sh

@@ -2,6 +2,8 @@
 
 set -ex
 
+source "$(dirname "${BASH_SOURCE[0]}")/common_utils.sh"
+
 if [ -n "${UBUNTU_VERSION}" ]; then
   apt update
   apt-get install -y clang doxygen git graphviz nodejs npm libtinfo5
@@ -13,8 +15,8 @@ chown -R jenkins pytorch
 
 pushd pytorch
 # Install all linter dependencies
-pip install -r requirements.txt
-lintrunner init
+pip_install -r requirements.txt
+conda_run lintrunner init
 
 # Cache .lintbin directory as part of the Docker image
 cp -r .lintbin /tmp

.ci/docker/common/install_python.sh

@@ -1,18 +0,0 @@
-#!/bin/bash
-set -ex
-
-apt-get update
-# Use deadsnakes in case we need an older python version
-sudo add-apt-repository ppa:deadsnakes/ppa
-apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python3-pip python${PYTHON_VERSION}-venv
-
-# Use a venv because uv and some other package managers don't support --user install
-ln -s /usr/bin/python${PYTHON_VERSION} /usr/bin/python
-python -m venv /var/lib/jenkins/ci_env
-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

@@ -8,6 +8,10 @@ ver() {
 
 install_ubuntu() {
     apt-get update
+    if [[ $UBUNTU_VERSION == 18.04 ]]; then
+      # gpg-agent is not available by default on 18.04
+      apt-get install -y --no-install-recommends gpg-agent
+    fi
     if [[ $UBUNTU_VERSION == 20.04 ]]; then
       # gpg-agent is not available by default on 20.04
       apt-get install -y --no-install-recommends gpg-agent

.ci/docker/common/install_rocm_drm.sh

@@ -25,9 +25,7 @@ python3 -m pip install meson ninja
 ###########################
 ### clone repo
 ###########################
-# TEMPORARY FIX: https://gitlab.freedesktop.org/mesa/drm.git is down until 2025/03/22
-# GIT_SSL_NO_VERIFY=true git clone https://gitlab.freedesktop.org/mesa/drm.git
-GIT_SSL_NO_VERIFY=true git clone git://anongit.freedesktop.org/mesa/drm
+GIT_SSL_NO_VERIFY=true git clone https://gitlab.freedesktop.org/mesa/drm.git
 pushd drm
 
 ###########################

.ci/docker/common/install_swiftshader.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+
+set -ex
+
+[ -n "${SWIFTSHADER}" ]
+
+retry () {
+    $*  || (sleep 1 && $*) || (sleep 2 && $*) || (sleep 4 && $*) || (sleep 8 && $*)
+}
+
+_https_amazon_aws=https://ossci-android.s3.amazonaws.com
+
+# SwiftShader
+_swiftshader_dir=/var/lib/jenkins/swiftshader
+_swiftshader_file_targz=swiftshader-abe07b943-prebuilt.tar.gz
+mkdir -p $_swiftshader_dir
+_tmp_swiftshader_targz="/tmp/${_swiftshader_file_targz}"
+
+curl --silent --show-error --location --fail --retry 3 \
+  --output "${_tmp_swiftshader_targz}" "$_https_amazon_aws/${_swiftshader_file_targz}"
+
+tar -C "${_swiftshader_dir}" -xzf "${_tmp_swiftshader_targz}"
+
+export VK_ICD_FILENAMES="${_swiftshader_dir}/build/Linux/vk_swiftshader_icd.json"

.ci/docker/common/install_vulkan_sdk.sh

@@ -0,0 +1,24 @@
+#!/bin/bash
+
+set -ex
+
+[ -n "${VULKAN_SDK_VERSION}" ]
+
+retry () {
+    $*  || (sleep 1 && $*) || (sleep 2 && $*) || (sleep 4 && $*) || (sleep 8 && $*)
+}
+
+_vulkansdk_dir=/var/lib/jenkins/vulkansdk
+_tmp_vulkansdk_targz=/tmp/vulkansdk.tar.gz
+
+curl \
+  --silent \
+  --show-error \
+  --location \
+  --fail \
+  --retry 3 \
+  --output "${_tmp_vulkansdk_targz}" "https://ossci-android.s3.amazonaws.com/vulkansdk-linux-x86_64-${VULKAN_SDK_VERSION}.tar.gz"
+
+mkdir -p "${_vulkansdk_dir}"
+tar -C "${_vulkansdk_dir}" -xzf "${_tmp_vulkansdk_targz}" --strip-components 1
+rm -rf "${_tmp_vulkansdk_targz}"

.ci/docker/common/install_xpu.sh

@@ -47,6 +47,9 @@ function install_ubuntu() {
     # Development Packages
     apt-get install -y libigc-dev intel-igc-cm libigdfcl-dev libigfxcmrt-dev level-zero-dev
     # Install Intel Support Packages
+    if [[ "$XPU_VERSION" == "2025.0" ]]; then
+        XPU_PACKAGES="${XPU_PACKAGES} intel-oneapi-dnnl=2025.0.1-6"
+    fi
     apt-get install -y ${XPU_PACKAGES}
 
     # Cleanup
@@ -82,6 +85,9 @@ gpgkey=https://yum.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.
 EOF
 
     # Install Intel Support Packages
+    if [[ "$XPU_VERSION" == "2025.0" ]]; then
+        XPU_PACKAGES="${XPU_PACKAGES} intel-oneapi-dnnl-2025.0.1-6"
+    fi
     yum install -y ${XPU_PACKAGES}
     # The xpu-smi packages
     dnf install -y xpu-smi

.ci/docker/linter-cuda/Dockerfile

@@ -18,14 +18,15 @@ COPY ./common/install_user.sh install_user.sh
 RUN bash ./install_user.sh && rm install_user.sh
 
 # Install conda and other packages (e.g., numpy, pytest)
-ARG PYTHON_VERSION
-ARG PIP_CMAKE
-# Put venv into the env vars so users don't need to activate it
-ENV PATH /var/lib/jenkins/ci_env/bin:$PATH
-ENV VIRTUAL_ENV /var/lib/jenkins/ci_env
-COPY requirements-ci.txt /opt/requirements-ci.txt
-COPY ./common/install_python.sh install_python.sh
-RUN bash ./install_python.sh && rm install_python.sh /opt/requirements-ci.txt
+ARG ANACONDA_PYTHON_VERSION
+ARG CONDA_CMAKE
+ENV ANACONDA_PYTHON_VERSION=$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
+COPY ./common/install_conda.sh install_conda.sh
+COPY ./common/common_utils.sh common_utils.sh
+COPY ./common/install_magma_conda.sh install_magma_conda.sh
+RUN bash ./install_conda.sh && rm install_conda.sh install_magma_conda.sh common_utils.sh /opt/conda/requirements-ci.txt
 
 # Install cuda and cudnn
 ARG CUDA_VERSION
@@ -36,10 +37,9 @@ ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:$PATH
 
 # Note that Docker build forbids copying file outside the build context
 COPY ./common/install_linter.sh install_linter.sh
+COPY ./common/common_utils.sh common_utils.sh
 RUN bash ./install_linter.sh
-RUN rm install_linter.sh
-
-RUN chown -R jenkins:jenkins /var/lib/jenkins/ci_env
+RUN rm install_linter.sh common_utils.sh
 
 USER jenkins
 CMD ["bash"]

.ci/docker/linter/Dockerfile

@@ -15,18 +15,20 @@ COPY ./common/install_user.sh install_user.sh
 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:$PATH
-ENV VIRTUAL_ENV /var/lib/jenkins/ci_env
-COPY requirements-ci.txt /opt/requirements-ci.txt
-COPY ./common/install_python.sh install_python.sh
-RUN bash ./install_python.sh && rm install_python.sh /opt/requirements-ci.txt
+ARG ANACONDA_PYTHON_VERSION
+ARG CONDA_CMAKE
+ENV ANACONDA_PYTHON_VERSION=$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
+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
 
 # Note that Docker build forbids copying file outside the build context
 COPY ./common/install_linter.sh install_linter.sh
+COPY ./common/common_utils.sh common_utils.sh
 RUN bash ./install_linter.sh
-RUN rm install_linter.sh
+RUN rm install_linter.sh common_utils.sh
 
 USER jenkins
 CMD ["bash"]

.ci/docker/manywheel/Dockerfile_2_28

@@ -60,11 +60,6 @@ FROM base as libpng
 ADD ./common/install_libpng.sh install_libpng.sh
 RUN bash ./install_libpng.sh && rm install_libpng.sh
 
-FROM base as cache
-# Install sccache
-ADD ./common/install_cache.sh install_cache.sh
-RUN bash ./install_cache.sh && rm install_cache.sh
-
 FROM ${GPU_IMAGE} as common
 ARG DEVTOOLSET_VERSION=11
 ENV LC_ALL en_US.UTF-8
@@ -117,10 +112,6 @@ COPY --from=libpng             /usr/local/include/libpng*            /usr/local/
 COPY --from=libpng             /usr/local/lib/libpng*                /usr/local/lib/
 COPY --from=libpng             /usr/local/lib/pkgconfig              /usr/local/lib/pkgconfig
 COPY --from=jni                /usr/local/include/jni.h              /usr/local/include/jni.h
-COPY --from=cache              /opt/cache                            /opt/cache
-
-# Ensure sccache is included in the path
-ENV PATH /opt/cache/bin:$PATH
 
 FROM common as cpu_final
 ARG BASE_CUDA_VERSION=11.8

.ci/docker/manywheel/Dockerfile_2_28_aarch64

@@ -59,11 +59,6 @@ FROM base as openblas
 ADD ./common/install_openblas.sh install_openblas.sh
 RUN bash ./install_openblas.sh && rm install_openblas.sh
 
-FROM base as cache
-# Install sccache
-ADD ./common/install_cache.sh install_cache.sh
-RUN bash ./install_cache.sh && rm install_cache.sh
-
 FROM base as final
 
 # remove unncessary python versions
@@ -72,8 +67,4 @@ 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/
-COPY --from=cache              /opt/cache                            /opt/cache
-
 ENV LD_LIBRARY_PATH=/opt/OpenBLAS/lib:$LD_LIBRARY_PATH
-# Ensure sccache is included in the path
-ENV PATH /opt/cache/bin:$PATH

.ci/docker/manywheel/Dockerfile_s390x

@@ -42,7 +42,6 @@ RUN yum install -y \
   llvm-devel \
   libzstd-devel \
   python3.12-devel \
-  python3.12-test \
   python3.12-setuptools \
   python3.12-pip \
   python3-virtualenv \
@@ -86,20 +85,11 @@ ADD ./common/install_cpython.sh /build_scripts/install_cpython.sh
 ENV SSL_CERT_FILE=
 RUN bash build_scripts/build.sh && rm -r build_scripts
 
-FROM base as cache
-# Install sccache
-ADD ./common/install_cache.sh install_cache.sh
-RUN bash ./install_cache.sh && rm install_cache.sh
-
 FROM base as final
 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=patchelf           /usr/local/bin/patchelf               /usr/local/bin/patchelf
-COPY --from=cache              /opt/cache                            /opt/cache
-
-# Ensure sccache is included in the path
-ENV PATH /opt/cache/bin:$PATH
 
 RUN alternatives --set python /usr/bin/python3.12
 RUN alternatives --set python3 /usr/bin/python3.12
@@ -111,33 +101,24 @@ CMD ["/bin/bash"]
 
 # install test dependencies:
 # - grpcio requires system openssl, bundled crypto fails to build
+# - ml_dtypes 0.4.0 requires some fixes provided in later commits to build
 RUN dnf install -y \
   protobuf-devel \
   protobuf-c-devel \
   protobuf-lite-devel \
-  hdf5-devel \
-  python3-h5py \
-  git
+  wget \
+  patch
 
-RUN env GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=True pip3 install grpcio
-
-# cmake-3.28.0 from pip for onnxruntime
-RUN python3 -mpip install cmake==3.28.0
-
-# build onnxruntime 1.21.0 from sources.
-# it is not possible to build it from sources using pip,
-# so just build it from upstream repository.
-# h5py is dependency of onnxruntime_training.
-# h5py==3.11.0 builds with hdf5-devel 1.10.5 from repository.
-# install newest flatbuffers version first:
-# for some reason old version is getting pulled in otherwise.
-# packaging package is required for onnxruntime wheel build.
-RUN pip3 install flatbuffers && \
-  pip3 install h5py==3.11.0 && \
-  pip3 install packaging && \
-  git clone https://github.com/microsoft/onnxruntime && \
-  cd onnxruntime && git checkout v1.21.0 && \
+RUN env GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=True pip3 install grpcio==1.65.4
+RUN cd ~ && \
+  git clone https://github.com/jax-ml/ml_dtypes && \
+  cd ml_dtypes && \
+  git checkout v0.4.0 && \
   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 && \
-  pip3 install ./build/Linux/Release/dist/onnxruntime_training-*.whl && \
-  cd .. && /bin/rm -rf ./onnxruntime
+  wget https://github.com/jax-ml/ml_dtypes/commit/b969f76914d6b30676721bc92bf0f6021a0d1321.patch && \
+  wget https://github.com/jax-ml/ml_dtypes/commit/d4e6d035ecda073eab8bcf60f4eef572ee7087e6.patch && \
+  patch -p1 < b969f76914d6b30676721bc92bf0f6021a0d1321.patch && \
+  patch -p1 < d4e6d035ecda073eab8bcf60f4eef572ee7087e6.patch && \
+  python3 setup.py bdist_wheel && \
+  pip3 install dist/*.whl && \
+  rm -rf ml_dtypes

.ci/docker/requirements-ci.txt

@@ -41,14 +41,11 @@ fbscribelogger==0.1.7
 #Pinned versions: 0.1.6
 #test that import:
 
-flatbuffers==2.0 ; platform_machine != "s390x"
+flatbuffers==2.0
 #Description: cross platform serialization library
 #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
@@ -105,10 +102,10 @@ networkx==2.8.8
 #Pinned versions: 2.8.8
 #test that import: functorch
 
-ninja==1.11.1.3
-#Description: build system. Used in some tests. Used in build to generate build
-#time tracing information
-#Pinned versions: 1.11.1.3
+#ninja
+#Description: build system.  Note that it install from
+#here breaks things so it is commented out
+#Pinned versions: 1.10.0.post1
 #test that import: run_test.py, test_cpp_extensions_aot.py,test_determination.py
 
 numba==0.49.0 ; python_version < "3.9"
@@ -368,6 +365,7 @@ PyYAML
 pyzstd
 setuptools
 
+ninja==1.11.1 ; platform_machine == "aarch64"
 scons==4.5.2 ; platform_machine == "aarch64"
 
 pulp==2.9.0 ; python_version >= "3.8"

.ci/docker/triton_version.txt

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

.ci/docker/ubuntu-cuda/Dockerfile

@@ -50,6 +50,13 @@ RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
 RUN rm install_protobuf.sh
 ENV INSTALLED_PROTOBUF ${PROTOBUF}
 
+# (optional) Install database packages like LMDB and LevelDB
+ARG DB
+COPY ./common/install_db.sh install_db.sh
+RUN if [ -n "${DB}" ]; then bash ./install_db.sh; fi
+RUN rm install_db.sh
+ENV INSTALLED_DB ${DB}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./

.ci/docker/ubuntu-rocm/Dockerfile

@@ -50,6 +50,13 @@ RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
 RUN rm install_protobuf.sh
 ENV INSTALLED_PROTOBUF ${PROTOBUF}
 
+# (optional) Install database packages like LMDB and LevelDB
+ARG DB
+COPY ./common/install_db.sh install_db.sh
+RUN if [ -n "${DB}" ]; then bash ./install_db.sh; fi
+RUN rm install_db.sh
+ENV INSTALLED_DB ${DB}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./

.ci/docker/ubuntu-xpu/Dockerfile

@@ -77,6 +77,13 @@ COPY triton_version.txt triton_version.txt
 RUN if [ -n "${TRITON}" ]; then bash ./install_triton.sh; fi
 RUN rm install_triton.sh common_utils.sh triton-xpu.txt triton_version.txt
 
+# (optional) Install database packages like LMDB and LevelDB
+ARG DB
+COPY ./common/install_db.sh install_db.sh
+RUN if [ -n "${DB}" ]; then bash ./install_db.sh; fi
+RUN rm install_db.sh
+ENV INSTALLED_DB ${DB}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./

.ci/docker/ubuntu/Dockerfile

@@ -1,12 +1,11 @@
 ARG UBUNTU_VERSION
 
-# ========================================
-# Builder stage
-# ========================================
-FROM ubuntu:${UBUNTU_VERSION} AS builder
+FROM ubuntu:${UBUNTU_VERSION}
 
 ARG UBUNTU_VERSION
+
 ENV DEBIAN_FRONTEND noninteractive
+
 ARG CLANG_VERSION
 
 # Install common dependencies (so that this step can be cached separately)
@@ -48,7 +47,7 @@ RUN bash ./install_gcc.sh && rm install_gcc.sh
 
 # Install lcov for C++ code coverage
 COPY ./common/install_lcov.sh install_lcov.sh
-RUN bash ./install_lcov.sh && rm install_lcov.sh
+RUN  bash ./install_lcov.sh && rm install_lcov.sh
 
 # Install cuda and cudnn
 ARG CUDA_VERSION
@@ -75,6 +74,13 @@ RUN if [ -n "${PROTOBUF}" ]; then bash ./install_protobuf.sh; fi
 RUN rm install_protobuf.sh
 ENV INSTALLED_PROTOBUF ${PROTOBUF}
 
+# (optional) Install database packages like LMDB and LevelDB
+ARG DB
+COPY ./common/install_db.sh install_db.sh
+RUN if [ -n "${DB}" ]; then bash ./install_db.sh; fi
+RUN rm install_db.sh
+ENV INSTALLED_DB ${DB}
+
 # (optional) Install vision packages like OpenCV
 ARG VISION
 COPY ./common/install_vision.sh ./common/cache_vision_models.sh ./common/common_utils.sh ./
@@ -82,6 +88,18 @@ RUN if [ -n "${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 Vulkan SDK
+ARG VULKAN_SDK_VERSION
+COPY ./common/install_vulkan_sdk.sh install_vulkan_sdk.sh
+RUN if [ -n "${VULKAN_SDK_VERSION}" ]; then bash ./install_vulkan_sdk.sh; fi
+RUN rm install_vulkan_sdk.sh
+
+# (optional) Install swiftshader
+ARG SWIFTSHADER
+COPY ./common/install_swiftshader.sh install_swiftshader.sh
+RUN if [ -n "${SWIFTSHADER}" ]; then bash ./install_swiftshader.sh; fi
+RUN rm install_swiftshader.sh
+
 # (optional) Install non-default CMake version
 ARG CMAKE_VERSION
 COPY ./common/install_cmake.sh install_cmake.sh
@@ -170,76 +188,19 @@ COPY ./common/install_openmpi.sh install_openmpi.sh
 RUN if [ -n "${CUDA_VERSION}" ]; then bash install_openmpi.sh; fi
 RUN rm install_openmpi.sh
 
+# Include BUILD_ENVIRONMENT environment variable in image
+ARG BUILD_ENVIRONMENT
+ENV BUILD_ENVIRONMENT ${BUILD_ENVIRONMENT}
+
 # Install LLVM dev version (Defined in the pytorch/builder github repository)
 ARG SKIP_LLVM_SRC_BUILD_INSTALL
 COPY --from=pytorch/llvm:9.0.1 /opt/llvm /opt/llvm
 RUN if [ -n "${SKIP_LLVM_SRC_BUILD_INSTALL}" ]; then set -eu; rm -rf /opt/llvm; fi
 
-# ========================================
-# Final stage
-# ========================================
-FROM ubuntu:${UBUNTU_VERSION}
-
-# Pass all arguments from the command line to this stage
-ARG UBUNTU_VERSION
-ARG CLANG_VERSION
-ARG LLVMDEV
-ARG KATEX
-ARG ANACONDA_PYTHON_VERSION
-ARG CONDA_CMAKE
-ARG DOCS
-ARG GCC_VERSION
-ARG CUDA_VERSION
-ARG UCX_COMMIT
-ARG UCC_COMMIT
-ARG PROTOBUF
-ARG VISION
-ARG CMAKE_VERSION
-ARG NINJA_VERSION
-ARG INDUCTOR_BENCHMARKS
-ARG TRITON
-ARG TRITON_CPU
-ARG EXECUTORCH
-ARG HALIDE
-ARG ONNX
-ARG ACL
-ARG BUILD_ENVIRONMENT
-
-ENV DEBIAN_FRONTEND noninteractive
-ENV ANACONDA_PYTHON_VERSION=$ANACONDA_PYTHON_VERSION
-ENV PATH /opt/conda/envs/py_$ANACONDA_PYTHON_VERSION/bin:/opt/conda/bin:$PATH
-ENV DOCS=$DOCS
-ENV DESIRED_CUDA ${CUDA_VERSION}
-ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:/opt/cache/bin:$PATH
-ENV UCX_COMMIT $UCX_COMMIT
-ENV UCC_COMMIT $UCC_COMMIT
-ENV UCX_HOME /usr
-ENV UCC_HOME /usr
-ENV INSTALLED_PROTOBUF ${PROTOBUF}
-ENV INSTALLED_VISION ${VISION}
-ENV OPENSSL_ROOT_DIR /opt/openssl
-ENV OPENSSL_DIR /opt/openssl
-ENV INSTALLED_ACL ${ACL}
-ENV BUILD_ENVIRONMENT ${BUILD_ENVIRONMENT}
+# AWS specific CUDA build guidance
 ENV TORCH_CUDA_ARCH_LIST Maxwell
 ENV TORCH_NVCC_FLAGS "-Xfatbin -compress-all"
 ENV CUDA_PATH /usr/local/cuda
 
-# Copy essential system directories and files from builder stage
-COPY --from=builder /etc/apt /etc/apt
-COPY --from=builder /usr/local /usr/local
-COPY --from=builder /opt /opt
-COPY --from=builder /usr/include /usr/include
-COPY --from=builder /usr/lib /usr/lib
-COPY --from=builder /usr/bin /usr/bin
-COPY --from=builder /usr/share /usr/share
-
-# Copy the user setup
-COPY --from=builder /etc/passwd /etc/passwd
-COPY --from=builder /etc/group /etc/group
-COPY --from=builder /etc/shadow /etc/shadow
-COPY --from=builder /etc/sudoers.d /etc/sudoers.d
-
-# Create and switch to jenkins user
 USER jenkins
-CMD ["bash"]
+CMD ["bash"]

.ci/manywheel/build_common.sh

@@ -111,6 +111,12 @@ case ${DESIRED_PYTHON} in
     ;;
 esac
 
+if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    export _GLIBCXX_USE_CXX11_ABI=1
+else
+    export _GLIBCXX_USE_CXX11_ABI=0
+fi
+
 if [[ "$DESIRED_CUDA" == *"rocm"* ]]; then
     echo "Calling build_amd.py at $(date)"
     python tools/amd_build/build_amd.py
@@ -203,6 +209,12 @@ if [[ -n "$BUILD_PYTHONLESS" ]]; then
 
     mkdir -p /tmp/$LIBTORCH_HOUSE_DIR
 
+    if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+        LIBTORCH_ABI="cxx11-abi-"
+    else
+        LIBTORCH_ABI=
+    fi
+
     zip -rq /tmp/$LIBTORCH_HOUSE_DIR/libtorch-$LIBTORCH_ABI$LIBTORCH_VARIANT-$PYTORCH_BUILD_VERSION.zip libtorch
     cp /tmp/$LIBTORCH_HOUSE_DIR/libtorch-$LIBTORCH_ABI$LIBTORCH_VARIANT-$PYTORCH_BUILD_VERSION.zip \
        /tmp/$LIBTORCH_HOUSE_DIR/libtorch-$LIBTORCH_ABI$LIBTORCH_VARIANT-latest.zip
@@ -321,8 +333,8 @@ 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 $destpath)
-            # Keep the so number for XPU dependencies
-            elif [[ "$DESIRED_CUDA" == *"xpu"* ]]; then
+            # Keep the so number for XPU dependencies and libgomp.so.1 to avoid twice load
+            elif [[ "$DESIRED_CUDA" == *"xpu"* || "$filename" == "libgomp.so.1" ]]; then
                 patchedpath=$destpath
             else
                 patchedpath=$(fname_with_sha256 $destpath)

.ci/manywheel/build_libtorch.sh

@@ -95,6 +95,12 @@ python setup.py clean
 retry pip install -qr requirements.txt
 retry pip install -q numpy==2.0.1
 
+if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    export _GLIBCXX_USE_CXX11_ABI=1
+else
+    export _GLIBCXX_USE_CXX11_ABI=0
+fi
+
 if [[ "$DESIRED_CUDA" == *"rocm"* ]]; then
     echo "Calling build_amd.py at $(date)"
     python tools/amd_build/build_amd.py
@@ -163,6 +169,12 @@ fi
 
 )
 
+if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    LIBTORCH_ABI="cxx11-abi-"
+else
+    LIBTORCH_ABI=
+fi
+
 (
     set -x
 

.ci/pytorch/check_binary.sh

@@ -59,28 +59,30 @@ else
   export install_root="$(dirname $(which python))/../lib/python${py_dot}/site-packages/torch/"
 fi
 
-###############################################################################
-# Setup XPU ENV
-###############################################################################
-if [[ "$DESIRED_CUDA" == 'xpu' ]]; then
-  set +u
-  # Refer https://www.intel.com/content/www/us/en/developer/articles/tool/pytorch-prerequisites-for-intel-gpus.html
-  source /opt/intel/oneapi/compiler/latest/env/vars.sh
-  source /opt/intel/oneapi/pti/latest/env/vars.sh
-fi
-
 ###############################################################################
 # Check GCC ABI
 ###############################################################################
 
-# NOTE: As of https://github.com/pytorch/pytorch/issues/126551 we only produce
-#       wheels with cxx11-abi
+# NOTE [ Building libtorch with old vs. new gcc ABI ]
+#
+# Packages built with one version of ABI could not be linked against by client
+# C++ libraries that were compiled using the other version of ABI. Since both
+# gcc ABIs are still common in the wild, we need to support both ABIs. Currently:
+#
+# - All the nightlies built on CentOS 7 + devtoolset7 use the old gcc ABI.
+# - All the nightlies built on Ubuntu 16.04 + gcc 5.4 use the new gcc ABI.
 
 echo "Checking that the gcc ABI is what we expect"
 if [[ "$(uname)" != 'Darwin' ]]; then
   function is_expected() {
-    if [[ "$1" -gt 0 || "$1" == "ON " ]]; then
-      echo 1
+    if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* || "$DESIRED_CUDA" == *"rocm"* ]]; then
+      if [[ "$1" -gt 0 || "$1" == "ON " ]]; then
+        echo 1
+      fi
+    else
+      if [[ -z "$1" || "$1" == 0 || "$1" == "OFF" ]]; then
+        echo 1
+      fi
     fi
   }
 
@@ -196,11 +198,35 @@ setup_link_flags () {
 
 TEST_CODE_DIR="$(dirname $(realpath ${BASH_SOURCE[0]}))/test_example_code"
 build_and_run_example_cpp () {
+  if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    GLIBCXX_USE_CXX11_ABI=1
+  else
+    GLIBCXX_USE_CXX11_ABI=0
+  fi
   setup_link_flags
-  g++ ${TEST_CODE_DIR}/$1.cpp -I${install_root}/include -I${install_root}/include/torch/csrc/api/include -std=gnu++17 -L${install_root}/lib ${REF_LIB} ${ADDITIONAL_LINKER_FLAGS} -ltorch $TORCH_CPU_LINK_FLAGS $TORCH_CUDA_LINK_FLAGS $C10_LINK_FLAGS -o $1
+  g++ ${TEST_CODE_DIR}/$1.cpp -I${install_root}/include -I${install_root}/include/torch/csrc/api/include -D_GLIBCXX_USE_CXX11_ABI=$GLIBCXX_USE_CXX11_ABI -std=gnu++17 -L${install_root}/lib ${REF_LIB} ${ADDITIONAL_LINKER_FLAGS} -ltorch $TORCH_CPU_LINK_FLAGS $TORCH_CUDA_LINK_FLAGS $C10_LINK_FLAGS -o $1
   ./$1
 }
 
+build_example_cpp_with_incorrect_abi () {
+  if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    GLIBCXX_USE_CXX11_ABI=0
+  else
+    GLIBCXX_USE_CXX11_ABI=1
+  fi
+  set +e
+  setup_link_flags
+  g++ ${TEST_CODE_DIR}/$1.cpp -I${install_root}/include -I${install_root}/include/torch/csrc/api/include -D_GLIBCXX_USE_CXX11_ABI=$GLIBCXX_USE_CXX11_ABI -std=gnu++17 -L${install_root}/lib ${REF_LIB} ${ADDITIONAL_LINKER_FLAGS} -ltorch $TORCH_CPU_LINK_FLAGS $TORCH_CUDA_LINK_FLAGS $C10_LINK_FLAGS -o $1
+  ERRCODE=$?
+  set -e
+  if [ "$ERRCODE" -eq "0" ]; then
+    echo "Building example with incorrect ABI didn't throw error. Aborting."
+    exit 1
+  else
+    echo "Building example with incorrect ABI throws expected error. Proceeding."
+  fi
+}
+
 ###############################################################################
 # Check simple Python/C++ calls
 ###############################################################################
@@ -210,6 +236,11 @@ if [[ "$PACKAGE_TYPE" == 'libtorch' ]]; then
     export LD_LIBRARY_PATH=/usr/local/cuda/lib64
   fi
   build_and_run_example_cpp simple-torch-test
+  # `_GLIBCXX_USE_CXX11_ABI` is always ignored by gcc in devtoolset7, so we test
+  # the expected failure case for Ubuntu 16.04 + gcc 5.4 only.
+  if [[ "$DESIRED_DEVTOOLSET" == *"cxx11-abi"* ]]; then
+    build_example_cpp_with_incorrect_abi simple-torch-test
+  fi
 else
   pushd /tmp
   python -c 'import torch'

.ci/pytorch/common_utils.sh

@@ -202,7 +202,7 @@ function install_torchrec_and_fbgemm() {
 
 function clone_pytorch_xla() {
   if [[ ! -d ./xla ]]; then
-    git clone --recursive --quiet https://github.com/pytorch/xla.git
+    git clone --recursive -b r2.7 https://github.com/pytorch/xla.git
     pushd xla
     # pin the xla hash so that we don't get broken by changes to xla
     git checkout "$(cat ../.github/ci_commit_pins/xla.txt)"

.ci/pytorch/install_cache_xla.sh

@@ -1,31 +1,50 @@
 #!/bin/bash
 
 # Script for installing sccache on the xla build job, which uses xla's docker
-# 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.
+# 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.
 
 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
-  # 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"
+  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
   chmod a+x "/tmp/cache/bin/$1"
 }
 

.ci/pytorch/smoke_test/check_binary_symbols.py

@@ -121,9 +121,9 @@ def main() -> None:
         else:
             install_root = Path(distutils.sysconfig.get_python_lib()) / "torch"
 
-    libtorch_cpu_path = str(install_root / "lib" / "libtorch_cpu.so")
-    # NOTE: All binaries are built with cxx11abi now
-    check_lib_symbols_for_abi_correctness(libtorch_cpu_path, False)
+    libtorch_cpu_path = install_root / "lib" / "libtorch_cpu.so"
+    pre_cxx11_abi = "cxx11-abi" not in os.getenv("DESIRED_DEVTOOLSET", "")
+    check_lib_symbols_for_abi_correctness(libtorch_cpu_path, pre_cxx11_abi)
 
 
 if __name__ == "__main__":

.ci/pytorch/smoke_test/check_gomp.py

@@ -0,0 +1,74 @@
+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

@@ -1173,8 +1173,9 @@ build_xla() {
   apply_patches
   SITE_PACKAGES="$(python -c 'from distutils.sysconfig import get_python_lib; print(get_python_lib())')"
   # These functions are defined in .circleci/common.sh in pytorch/xla repo
-  retry install_deps_pytorch_xla $XLA_DIR $USE_CACHE
+  retry install_pre_deps_pytorch_xla $XLA_DIR $USE_CACHE
   CMAKE_PREFIX_PATH="${SITE_PACKAGES}/torch:${CMAKE_PREFIX_PATH}" XLA_SANDBOX_BUILD=1 build_torch_xla $XLA_DIR
+  retry install_post_deps_pytorch_xla
   assert_git_not_dirty
 }
 
@@ -1474,8 +1475,7 @@ test_executorch() {
   pushd /executorch
 
   export PYTHON_EXECUTABLE=python
-  export EXECUTORCH_BUILD_PYBIND=ON
-  export CMAKE_ARGS="-DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON"
+  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
@@ -1619,7 +1619,6 @@ elif [[ "${TEST_CONFIG}" == *inductor_cpp_wrapper* ]]; then
   install_torchvision
   checkout_install_torchbench hf_T5 llama moco
   PYTHONPATH=$(pwd)/torchbench test_inductor_cpp_wrapper_shard "$SHARD_NUMBER"
-  test_inductor_aoti
 elif [[ "${TEST_CONFIG}" == *inductor* ]]; then
   install_torchvision
   test_inductor_shard "${SHARD_NUMBER}"

.ci/pytorch/windows/internal/smoke_test.bat

@@ -128,6 +128,7 @@ goto end
 :libtorch
 echo "install and test libtorch"
 
+if "%VC_YEAR%" == "2019" powershell internal\vs2019_install.ps1
 if "%VC_YEAR%" == "2022" powershell internal\vs2022_install.ps1
 
 if ERRORLEVEL 1 exit /b 1
@@ -139,6 +140,10 @@ pushd tmp\libtorch
 
 set VC_VERSION_LOWER=17
 set VC_VERSION_UPPER=18
+IF "%VC_YEAR%" == "2019" (
+    set VC_VERSION_LOWER=16
+    set VC_VERSION_UPPER=17
+)
 
 for /f "usebackq tokens=*" %%i in (`"%ProgramFiles(x86)%\Microsoft Visual Studio\Installer\vswhere.exe" -legacy -products * -version [%VC_VERSION_LOWER%^,%VC_VERSION_UPPER%^) -property installationPath`) do (
     if exist "%%i" if exist "%%i\VC\Auxiliary\Build\vcvarsall.bat" (

.ci/pytorch/windows/internal/static_lib_test.bat

@@ -70,6 +70,7 @@ echo "install and test libtorch"
 pip install cmake
 echo "installing cmake"
 
+if "%VC_YEAR%" == "2019" powershell internal\vs2019_install.ps1
 if "%VC_YEAR%" == "2022" powershell internal\vs2022_install.ps1
 
 if ERRORLEVEL 1 exit /b 1
@@ -82,6 +83,10 @@ pushd tmp\libtorch
 
 set VC_VERSION_LOWER=17
 set VC_VERSION_UPPER=18
+IF "%VC_YEAR%" == "2019" (
+    set VC_VERSION_LOWER=16
+    set VC_VERSION_UPPER=17
+)
 
 for /f "usebackq tokens=*" %%i in (`"%ProgramFiles(x86)%\Microsoft Visual Studio\Installer\vswhere.exe" -legacy -products * -version [%VC_VERSION_LOWER%^,%VC_VERSION_UPPER%^) -property installationPath`) do (
     if exist "%%i" if exist "%%i\VC\Auxiliary\Build\vcvarsall.bat" (

.ci/pytorch/windows/internal/vc_install_helper.bat

@@ -1,8 +1,12 @@
+if "%VC_YEAR%" == "2019" powershell windows/internal/vs2019_install.ps1
 if "%VC_YEAR%" == "2022" powershell windows/internal/vs2022_install.ps1
 
 set VC_VERSION_LOWER=17
 set VC_VERSION_UPPER=18
-
+if "%VC_YEAR%" == "2019" (
+    set VC_VERSION_LOWER=16
+    set VC_VERSION_UPPER=17
+)
 
 for /f "usebackq tokens=*" %%i in (`"%ProgramFiles(x86)%\Microsoft Visual Studio\Installer\vswhere.exe"  -products Microsoft.VisualStudio.Product.BuildTools -version [%VC_VERSION_LOWER%^,%VC_VERSION_UPPER%^) -property installationPath`) do (
     if exist "%%i" if exist "%%i\VC\Auxiliary\Build\vcvarsall.bat" (

.ci/pytorch/windows/internal/vs2019_install.ps1

@@ -0,0 +1,48 @@
+# https://developercommunity.visualstudio.com/t/install-specific-version-of-vs-component/1142479
+# https://docs.microsoft.com/en-us/visualstudio/releases/2019/history#release-dates-and-build-numbers
+
+# 16.8.6 BuildTools
+$VS_DOWNLOAD_LINK = "https://ossci-windows.s3.us-east-1.amazonaws.com/vs16.8.6_BuildTools.exe"
+$COLLECT_DOWNLOAD_LINK = "https://aka.ms/vscollect.exe"
+$VS_INSTALL_ARGS = @("--nocache","--quiet","--wait", "--add Microsoft.VisualStudio.Workload.VCTools",
+                                                     "--add Microsoft.Component.MSBuild",
+                                                     "--add Microsoft.VisualStudio.Component.Roslyn.Compiler",
+                                                     "--add Microsoft.VisualStudio.Component.TextTemplating",
+                                                     "--add Microsoft.VisualStudio.Component.VC.CoreIde",
+                                                     "--add Microsoft.VisualStudio.Component.VC.Redist.14.Latest",
+                                                     "--add Microsoft.VisualStudio.ComponentGroup.NativeDesktop.Core",
+                                                     "--add Microsoft.VisualStudio.Component.VC.Tools.x86.x64",
+                                                     "--add Microsoft.VisualStudio.ComponentGroup.NativeDesktop.Win81")
+
+curl.exe --retry 3 -kL $VS_DOWNLOAD_LINK --output vs_installer.exe
+if ($LASTEXITCODE -ne 0) {
+    echo "Download of the VS 2019 Version 16.8.5 installer failed"
+    exit 1
+}
+
+if (Test-Path "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe") {
+    $existingPath = & "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe" -products "Microsoft.VisualStudio.Product.BuildTools" -version "[16, 17)" -property installationPath
+    if ($existingPath -ne $null) {
+        if (!${env:CIRCLECI}) {
+            echo "Found correctly versioned existing BuildTools installation in $existingPath"
+            exit 0
+        }
+        echo "Found existing BuildTools installation in $existingPath, keeping it"
+    }
+}
+
+$process = Start-Process "${PWD}\vs_installer.exe" -ArgumentList $VS_INSTALL_ARGS -NoNewWindow -Wait -PassThru
+Remove-Item -Path vs_installer.exe -Force
+$exitCode = $process.ExitCode
+if (($exitCode -ne 0) -and ($exitCode -ne 3010)) {
+    echo "VS 2019 installer exited with code $exitCode, which should be one of [0, 3010]."
+    curl.exe --retry 3 -kL $COLLECT_DOWNLOAD_LINK --output Collect.exe
+    if ($LASTEXITCODE -ne 0) {
+        echo "Download of the VS Collect tool failed."
+        exit 1
+    }
+    Start-Process "${PWD}\Collect.exe" -NoNewWindow -Wait -PassThru
+    New-Item -Path "C:\w\build-results" -ItemType "directory" -Force
+    Copy-Item -Path "C:\Users\${env:USERNAME}\AppData\Local\Temp\vslogs.zip" -Destination "C:\w\build-results\"
+    exit 1
+}

.circleci/scripts/binary_linux_test.sh

@@ -92,6 +92,11 @@ fi
 if [[ "\$GPU_ARCH_TYPE" != *s390x* && "\$GPU_ARCH_TYPE" != *xpu* && "\$GPU_ARCH_TYPE" != *rocm*  && "$PACKAGE_TYPE" != libtorch ]]; then
   # Exclude s390, xpu, rocm and libtorch builds from smoke testing
   python /pytorch/.ci/pytorch/smoke_test/smoke_test.py --package=torchonly --torch-compile-check disabled
+
+  if [[ "\$GPU_ARCH_TYPE" != *cpu-aarch64* ]]; then
+    # test for issue https://github.com/pytorch/pytorch/issues/149422
+    python /pytorch/.ci/pytorch/smoke_test/check_gomp.py
+  fi
 fi
 
 # Clean temp files

.circleci/scripts/binary_upload.sh

@@ -55,16 +55,12 @@ s3_upload() {
     s3_upload_dir="${s3_root_dir}/${UPLOAD_SUBFOLDER}/"
   fi
   (
-    cache_control_flag=""
-    if [[ "${UPLOAD_CHANNEL}" = "test" ]]; then
-      cache_control_flag="--cache-control='no-cache,no-store,must-revalidate'"
-    fi
     for pkg in ${PKG_DIR}/*.${extension}; do
       (
         set -x
         shm_id=$(sha256sum "${pkg}" | awk '{print $1}')
         ${AWS_S3_CP} --no-progress --acl public-read "${pkg}" "${s3_upload_dir}" \
-          --metadata "checksum-sha256=${shm_id}" ${cache_control_flag}
+          --metadata "checksum-sha256=${shm_id}"
       )
     done
   )

.circleci/scripts/binary_windows_build.sh

@@ -8,9 +8,10 @@ export CUDA_VERSION="${DESIRED_CUDA/cu/}"
 export USE_SCCACHE=1
 export SCCACHE_BUCKET=ossci-compiler-cache
 export SCCACHE_IGNORE_SERVER_IO_ERROR=1
-export VC_YEAR=2022
+export VC_YEAR=2019
 
 if [[ "$DESIRED_CUDA" == 'xpu' ]]; then
+    export VC_YEAR=2022
     export USE_SCCACHE=0
     export XPU_VERSION=2025.0
     export XPU_ENABLE_KINETO=1

.circleci/scripts/binary_windows_test.sh

@@ -4,9 +4,10 @@ set -eux -o pipefail
 source "${BINARY_ENV_FILE:-/c/w/env}"
 
 export CUDA_VERSION="${DESIRED_CUDA/cu/}"
-export VC_YEAR=2022
+export VC_YEAR=2019
 
 if [[ "$DESIRED_CUDA" == 'xpu' ]]; then
+    export VC_YEAR=2022
     export XPU_VERSION=2025.0
 fi
 

.clang-tidy

@@ -48,6 +48,7 @@ misc-*,
 -misc-no-recursion,
 -misc-non-private-member-variables-in-classes,
 -misc-unused-using-decls,
+-misc-use-internal-linkage,
 modernize-*,
 -modernize-macro-to-enum,
 -modernize-return-braced-init-list,

.github/ISSUE_TEMPLATE/disable-ci-jobs.md

@@ -5,7 +5,7 @@ title: "DISABLED [WORKFLOW_NAME] / [PLATFORM_NAME] / [JOB_NAME]"
 labels: "module: ci"
 ---
 
-> For example, DISABLED pull / win-vs2022-cpu-py3 / test (default). Once
+> For example, DISABLED pull / win-vs2019-cpu-py3 / test (default). Once
 > created, the job will be disabled within 15 minutes. You can check the
 > list of disabled jobs at https://ossci-metrics.s3.amazonaws.com/disabled-jobs.json
 

.github/actionlint.yaml

@@ -3,9 +3,6 @@ self-hosted-runner:
     # GitHub hosted runner that actionlint doesn't recognize because actionlint version (1.6.21) is too old
     - ubuntu-24.04
     # GitHub hosted x86 Linux runners
-    # TODO: Cleanup mentions of linux.20_04 when upgrade to linux.24_04 is complete
-    - linux.20_04.4x
-    - linux.20_04.16x
     - linux.24_04.4x
     - linux.24_04.16x
     # Organization-wide AWS Linux Runners
@@ -52,7 +49,6 @@ self-hosted-runner:
     - linux.rocm.gpu
     - linux.rocm.gpu.2
     - linux.rocm.gpu.4
-    - rocm-docker
     # Repo-specific Apple hosted  runners
     - macos-m1-ultra
     - macos-m2-14

.github/actions/checkout-pytorch/action.yml

@@ -23,44 +23,9 @@ runs:
       id: check_container_runner
       run: echo "IN_CONTAINER_RUNNER=$(if [ -f /.inarc ] || [ -f /.incontainer ]; then echo true ; else echo false; fi)" >> "$GITHUB_OUTPUT"
 
-    - name: Set up parallel fetch and clean workspace
-      id: first-clean
-      continue-on-error: true
+    - name: Clean workspace
       shell: bash
       if: ${{ steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' }}
-      env:
-        NO_SUDO: ${{ inputs.no-sudo }}
-      run: |
-        # Use all available CPUs for fetching
-        cd "${GITHUB_WORKSPACE}"
-        git config --global fetch.parallel 0
-        git config --global submodule.fetchJobs 0
-
-        # Clean workspace. The default checkout action should also do this, but
-        # do it here as well just in case
-        if [[ -d .git ]]; then
-          if [ -z "${NO_SUDO}" ]; then
-            sudo git clean -ffdx
-          else
-            git clean -ffdx
-          fi
-        fi
-
-    - name: Checkout PyTorch
-      id: first-checkout-attempt
-      continue-on-error: true
-      uses: actions/checkout@v4
-      with:
-        ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
-        # --depth=1 for speed, manually fetch history and other refs as necessary
-        fetch-depth: ${{ inputs.fetch-depth }}
-        submodules: ${{ inputs.submodules }}
-        show-progress: false
-
-    - name: Clean workspace (try again)
-      if: ${{ steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' &&
-        (steps.first-clean.outcome != 'success' || steps.first-checkout-attempt.outcome != 'success') }}
-      shell: bash
       env:
         NO_SUDO: ${{ inputs.no-sudo }}
       run: |
@@ -75,11 +40,16 @@ runs:
         fi
         mkdir "${GITHUB_WORKSPACE}"
 
-    - name: Checkout PyTorch (try again)
+        # Use all available CPUs for fetching
+        cd "${GITHUB_WORKSPACE}"
+        git config --global fetch.parallel 0
+        git config --global submodule.fetchJobs 0
+
+    - name: Checkout PyTorch
       uses: actions/checkout@v4
-      if: ${{ steps.first-clean.outcome != 'success' || steps.first-checkout-attempt.outcome != 'success' }}
       with:
         ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+        # --depth=1 for speed, manually fetch history and other refs as necessary
         fetch-depth: ${{ inputs.fetch-depth }}
         submodules: ${{ inputs.submodules }}
         show-progress: false

.github/actions/test-pytorch-binary/action.yml

@@ -15,6 +15,7 @@ runs:
           -e BINARY_ENV_FILE \
           -e BUILD_ENVIRONMENT \
           -e DESIRED_CUDA \
+          -e DESIRED_DEVTOOLSET \
           -e DESIRED_PYTHON \
           -e GITHUB_ACTIONS \
           -e GPU_ARCH_TYPE \

.github/ci_commit_pins/audio.txt

@@ -1 +1 @@
-318bace01aebc1f82ae13d0d133fcf9fede73383
+c670ad81fda266b6598aeeef434583eb98197ae8

.github/ci_commit_pins/xla.txt

@@ -1 +1 @@
-b2b890e962f5fb6f481e5da2eb4a43bb990d0f1b
+r2.7

.github/scripts/filter_test_configs.py

@@ -39,9 +39,9 @@ SUPPORTED_PERIODICAL_MODES: dict[str, Callable[[Optional[str]], bool]] = {
 }
 
 # The link to the published list of disabled jobs
-DISABLED_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/disabled-jobs.json"
+DISABLED_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/disabled-jobs.json?versionId=n.FT07XR3dLMwOLBwmRNquyYSeGk8Het"
 # and unstable jobs
-UNSTABLE_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/unstable-jobs.json"
+UNSTABLE_JOBS_URL = "https://ossci-metrics.s3.amazonaws.com/unstable-jobs.json?versionId=.Ox7WAXa21I1PVqadHyPfhMRPhl0aCnD"
 
 # Some constants used to handle disabled and unstable jobs
 JOB_NAME_SEP = "/"

.github/scripts/generate_binary_build_matrix.py

@@ -34,6 +34,8 @@ ROCM_ARCHES = ["6.2.4", "6.3"]
 
 XPU_ARCHES = ["xpu"]
 
+CPU_CXX11_ABI_ARCH = ["cpu-cxx11-abi"]
+
 CPU_AARCH64_ARCH = ["cpu-aarch64"]
 
 CPU_S390X_ARCH = ["cpu-s390x"]
@@ -75,7 +77,7 @@ PYTORCH_EXTRA_INSTALL_REQUIREMENTS = {
         "nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | "
         "nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | "
         "nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | "
-        "nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | "
+        "nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | "
         "nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | "
         "nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | "
         "nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | "
@@ -144,6 +146,8 @@ def arch_type(arch_version: str) -> str:
         return "rocm"
     elif arch_version in XPU_ARCHES:
         return "xpu"
+    elif arch_version in CPU_CXX11_ABI_ARCH:
+        return "cpu-cxx11-abi"
     elif arch_version in CPU_AARCH64_ARCH:
         return "cpu-aarch64"
     elif arch_version in CPU_S390X_ARCH:
@@ -172,23 +176,31 @@ WHEEL_CONTAINER_IMAGES = {
     },
     "xpu": f"pytorch/manylinux2_28-builder:xpu-{DEFAULT_TAG}",
     "cpu": f"pytorch/manylinux2_28-builder:cpu-{DEFAULT_TAG}",
+    "cpu-cxx11-abi": f"pytorch/manylinuxcxx11-abi-builder:cpu-cxx11-abi-{DEFAULT_TAG}",
     "cpu-aarch64": f"pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-{DEFAULT_TAG}",
     "cpu-s390x": f"pytorch/manylinuxs390x-builder:cpu-s390x-{DEFAULT_TAG}",
 }
 
+CXX11_ABI = "cxx11-abi"
 RELEASE = "release"
 DEBUG = "debug"
 
-LIBTORCH_CONTAINER_IMAGES: dict[str, str] = {
+LIBTORCH_CONTAINER_IMAGES: dict[tuple[str, str], str] = {
     **{
-        gpu_arch: f"pytorch/libtorch-cxx11-builder:cuda{gpu_arch}-{DEFAULT_TAG}"
+        (
+            gpu_arch,
+            CXX11_ABI,
+        ): f"pytorch/libtorch-cxx11-builder:cuda{gpu_arch}-{DEFAULT_TAG}"
         for gpu_arch in CUDA_ARCHES
     },
     **{
-        gpu_arch: f"pytorch/libtorch-cxx11-builder:rocm{gpu_arch}-{DEFAULT_TAG}"
+        (
+            gpu_arch,
+            CXX11_ABI,
+        ): f"pytorch/libtorch-cxx11-builder:rocm{gpu_arch}-{DEFAULT_TAG}"
         for gpu_arch in ROCM_ARCHES
     },
-    "cpu": f"pytorch/libtorch-cxx11-builder:cpu-{DEFAULT_TAG}",
+    ("cpu", CXX11_ABI): f"pytorch/libtorch-cxx11-builder:cpu-{DEFAULT_TAG}",
 }
 
 FULL_PYTHON_VERSIONS = ["3.9", "3.10", "3.11", "3.12", "3.13", "3.13t"]
@@ -198,6 +210,7 @@ def translate_desired_cuda(gpu_arch_type: str, gpu_arch_version: str) -> str:
     return {
         "cpu": "cpu",
         "cpu-aarch64": "cpu",
+        "cpu-cxx11-abi": "cpu-cxx11-abi",
         "cpu-s390x": "cpu",
         "cuda": f"cu{gpu_arch_version.replace('.', '')}",
         "cuda-aarch64": f"cu{gpu_arch_version.replace('-aarch64', '').replace('.', '')}",
@@ -212,7 +225,7 @@ def list_without(in_list: list[str], without: list[str]) -> list[str]:
 
 def generate_libtorch_matrix(
     os: str,
-    release_type: str,
+    abi_version: str,
     arches: Optional[list[str]] = None,
     libtorch_variants: Optional[list[str]] = None,
 ) -> list[dict[str, str]]:
@@ -234,6 +247,9 @@ def generate_libtorch_matrix(
     ret: list[dict[str, str]] = []
     for arch_version in arches:
         for libtorch_variant in libtorch_variants:
+            # one of the values in the following list must be exactly
+            # CXX11_ABI, but the precise value of the other one doesn't
+            # matter
             gpu_arch_type = arch_type(arch_version)
             gpu_arch_version = "" if arch_version == "cpu" else arch_version
             # ROCm builds without-deps failed even in ROCm runners; skip for now
@@ -246,15 +262,20 @@ def generate_libtorch_matrix(
                     "desired_cuda": translate_desired_cuda(
                         gpu_arch_type, gpu_arch_version
                     ),
-                    "libtorch_config": release_type,
                     "libtorch_variant": libtorch_variant,
+                    "libtorch_config": abi_version
+                    if os in ("windows", "windows-arm64")
+                    else "",
+                    "devtoolset": abi_version
+                    if os not in ("windows", "windows-arm64")
+                    else "",
                     "container_image": (
-                        LIBTORCH_CONTAINER_IMAGES[arch_version]
+                        LIBTORCH_CONTAINER_IMAGES[(arch_version, abi_version)]
                         if os not in ("windows", "windows-arm64")
                         else ""
                     ),
                     "package_type": "libtorch",
-                    "build_name": f"libtorch-{gpu_arch_type}{gpu_arch_version}-{libtorch_variant}-{release_type}".replace(
+                    "build_name": f"libtorch-{gpu_arch_type}{gpu_arch_version}-{libtorch_variant}-{abi_version}".replace(
                         ".", "_"
                     ),
                 }
@@ -280,7 +301,7 @@ def generate_wheels_matrix(
         # Define default compute archivectures
         arches = ["cpu"]
         if os == "linux":
-            arches += CUDA_ARCHES + ROCM_ARCHES + XPU_ARCHES
+            arches += CPU_CXX11_ABI_ARCH + CUDA_ARCHES + ROCM_ARCHES + XPU_ARCHES
         elif os == "windows":
             arches += CUDA_ARCHES + XPU_ARCHES
         elif os == "linux-aarch64":
@@ -299,6 +320,7 @@ def generate_wheels_matrix(
             gpu_arch_version = (
                 ""
                 if arch_version == "cpu"
+                or arch_version == "cpu-cxx11-abi"
                 or arch_version == "cpu-aarch64"
                 or arch_version == "cpu-s390x"
                 or arch_version == "xpu"
@@ -333,6 +355,7 @@ def generate_wheels_matrix(
                         "gpu_arch_version": gpu_arch_version,
                         "desired_cuda": desired_cuda,
                         "use_split_build": "True" if use_split_build else "False",
+                        "devtoolset": "cxx11-abi",
                         "container_image": WHEEL_CONTAINER_IMAGES[arch_version],
                         "package_type": package_type,
                         "pytorch_extra_install_requirements": (
@@ -361,6 +384,7 @@ def generate_wheels_matrix(
                                 gpu_arch_type, gpu_arch_version
                             ),
                             "use_split_build": "True" if use_split_build else "False",
+                            "devtoolset": "",
                             "container_image": WHEEL_CONTAINER_IMAGES[arch_version],
                             "package_type": package_type,
                             "pytorch_extra_install_requirements": "",
@@ -379,6 +403,12 @@ def generate_wheels_matrix(
                             gpu_arch_type, gpu_arch_version
                         ),
                         "use_split_build": "True" if use_split_build else "False",
+                        "devtoolset": (
+                            "cxx11-abi"
+                            if (arch_version in ["cpu-cxx11-abi", "cpu-aarch64"])
+                            or os == "linux"
+                            else ""
+                        ),
                         "container_image": WHEEL_CONTAINER_IMAGES[arch_version],
                         "package_type": package_type,
                         "build_name": f"{package_type}-py{python_version}-{gpu_arch_type}{gpu_arch_version}".replace(
@@ -387,8 +417,6 @@ def generate_wheels_matrix(
                         "pytorch_extra_install_requirements": (
                             PYTORCH_EXTRA_INSTALL_REQUIREMENTS["xpu"]
                             if gpu_arch_type == "xpu"
-                            else PYTORCH_EXTRA_INSTALL_REQUIREMENTS[CUDA_STABLE]
-                            if os != "linux"
                             else ""
                         ),
                     }

.github/scripts/generate_ci_workflows.py

@@ -54,6 +54,7 @@ class BinaryBuildWorkflow:
 
     # Optional fields
     build_environment: str = ""
+    abi_version: str = ""
     ciflow_config: CIFlowConfig = field(default_factory=CIFlowConfig)
     is_scheduled: str = ""
     branches: str = "nightly"
@@ -63,7 +64,12 @@ class BinaryBuildWorkflow:
     use_split_build: bool = False
 
     def __post_init__(self) -> None:
-        self.build_environment = f"{self.os}-binary-{self.package_type}"
+        if self.abi_version:
+            self.build_environment = (
+                f"{self.os}-binary-{self.package_type}-{self.abi_version}"
+            )
+        else:
+            self.build_environment = f"{self.os}-binary-{self.package_type}"
         if self.use_split_build:
             # added to distinguish concurrency groups
             self.build_environment += "-split"
@@ -127,9 +133,10 @@ LINUX_BINARY_BUILD_WORFKLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.LINUX,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.CXX11_ABI,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.LINUX,
-            generate_binary_build_matrix.RELEASE,
+            generate_binary_build_matrix.CXX11_ABI,
             libtorch_variants=["shared-with-deps"],
         ),
         ciflow_config=CIFlowConfig(
@@ -169,9 +176,10 @@ LINUX_BINARY_SMOKE_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.LINUX,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.CXX11_ABI,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.LINUX,
-            generate_binary_build_matrix.RELEASE,
+            generate_binary_build_matrix.CXX11_ABI,
             arches=["cpu"],
             libtorch_variants=["shared-with-deps"],
         ),
@@ -194,6 +202,7 @@ WINDOWS_BINARY_BUILD_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.RELEASE,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS,
             generate_binary_build_matrix.RELEASE,
@@ -207,6 +216,7 @@ WINDOWS_BINARY_BUILD_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.DEBUG,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS,
             generate_binary_build_matrix.DEBUG,
@@ -223,6 +233,7 @@ WINDOWS_BINARY_SMOKE_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.RELEASE,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS,
             generate_binary_build_matrix.RELEASE,
@@ -237,6 +248,7 @@ WINDOWS_BINARY_SMOKE_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.DEBUG,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS,
             generate_binary_build_matrix.DEBUG,
@@ -267,6 +279,7 @@ WINDOWS_ARM64_BINARY_BUILD_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS_ARM64,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.RELEASE,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS_ARM64,
             generate_binary_build_matrix.RELEASE,
@@ -281,6 +294,7 @@ WINDOWS_ARM64_BINARY_BUILD_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.WINDOWS_ARM64,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.DEBUG,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.WINDOWS_ARM64,
             generate_binary_build_matrix.DEBUG,
@@ -298,9 +312,10 @@ MACOS_BINARY_BUILD_WORKFLOWS = [
     BinaryBuildWorkflow(
         os=OperatingSystem.MACOS_ARM64,
         package_type="libtorch",
+        abi_version=generate_binary_build_matrix.CXX11_ABI,
         build_configs=generate_binary_build_matrix.generate_libtorch_matrix(
             OperatingSystem.MACOS,
-            generate_binary_build_matrix.RELEASE,
+            generate_binary_build_matrix.CXX11_ABI,
             libtorch_variants=["shared-with-deps"],
         ),
         cross_compile_arm64=False,

.github/scripts/lintrunner.sh

@@ -1,6 +1,11 @@
 #!/usr/bin/env bash
 set -ex
 
+# The generic Linux job chooses to use base env, not the one setup by the image
+CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
+eval "$(command conda 'shell.bash' 'hook' 2> /dev/null)"
+conda activate "${CONDA_ENV}"
+
 # Use uv to speed up lintrunner init
 python3 -m pip install uv==0.1.45
 

.github/scripts/s390x-ci/self-hosted-builder/actions-runner.Dockerfile

@@ -5,50 +5,6 @@ FROM --platform=linux/amd64 docker.io/ubuntu:24.04 as ld-prefix
 ENV DEBIAN_FRONTEND=noninteractive
 RUN apt-get update && apt-get -y install ca-certificates libicu74 libssl3
 
-# Patched podman
-FROM --platform=linux/s390x docker.io/ubuntu:24.04 as podman
-ENV DEBIAN_FRONTEND=noninteractive
-RUN sed -i 's/^Types: deb$/Types: deb deb-src/' /etc/apt/sources.list.d/ubuntu.sources
-RUN apt-get update && \
-    apt-get install -y \
-        cmake \
-        curl \
-        devscripts \
-        dpkg-dev \
-        gdb \
-        less \
-        make \
-        python3 \
-        python3-pip \
-        quilt \
-        rsync \
-        software-properties-common \
-        stress-ng \
-        vim \
-        nano \
-        wget && \
-    apt-get build-dep -y podman && \
-    apt-get source podman
-
-COPY podman-patches/podman-25245.patch /tmp/podman-25245.patch
-COPY podman-patches/podman-25102-backport.patch /tmp/podman-25102-backport.patch
-
-# import and apply patches
-# patches:
-# https://github.com/containers/podman/pull/25102
-# https://github.com/containers/podman/pull/25245
-RUN cd /libpod-* && \
-    quilt import /tmp/podman-25245.patch && quilt push && \
-    quilt import /tmp/podman-25102-backport.patch && quilt push && \
-    dch -i "Fix podman deadlock and add option to clean up build leftovers" && \
-    /bin/rm /tmp/podman-25245.patch /tmp/podman-25102-backport.patch
-
-# build patched podman
-RUN cd /libpod-* && \
-    debuild -i -us -uc -b && \
-    /bin/rm /podman-remote_*.deb && \
-    mkdir /tmp/podman && cp -v /podman*.deb /tmp/podman
-
 # Main image.
 FROM --platform=linux/s390x docker.io/ubuntu:24.04
 
@@ -89,11 +45,7 @@ COPY fs/ /
 RUN chmod +x /usr/bin/actions-runner /usr/bin/entrypoint
 
 # install podman
-# RUN apt-get update && apt -y install podman podman-docker
-
-# install patched podman
-COPY --from=podman /tmp/podman /tmp/podman
-RUN apt-get update && apt -y install /tmp/podman/*.deb && /bin/rm -rfv /tmp/podman
+RUN apt -y install podman podman-docker
 
 # amd64 Github Actions Runner.
 RUN useradd -m actions-runner
@@ -113,7 +65,7 @@ RUN virtualenv --system-site-packages venv
 #
 COPY --chown=actions-runner:actions-runner manywheel-s390x.tar /home/actions-runner/manywheel-s390x.tar
 
-RUN curl -L https://github.com/actions/runner/releases/download/v2.322.0/actions-runner-linux-x64-2.322.0.tar.gz | tar -xz
+RUN curl -L https://github.com/actions/runner/releases/download/v2.317.0/actions-runner-linux-x64-2.317.0.tar.gz | tar -xz
 
 ENTRYPOINT ["/usr/bin/entrypoint"]
 CMD ["/usr/bin/actions-runner"]

.github/scripts/s390x-ci/self-hosted-builder/podman-patches/podman-25102-backport.patch

@@ -1,358 +0,0 @@
-diff --git a/cmd/podman/system/prune.go b/cmd/podman/system/prune.go
-index f7cf7b551..739f87cde 100644
---- a/cmd/podman/system/prune.go
-+++ b/cmd/podman/system/prune.go
-@@ -48,6 +48,7 @@ func init() {
- 	flags.BoolVarP(&force, "force", "f", false, "Do not prompt for confirmation.  The default is false")
- 	flags.BoolVarP(&pruneOptions.All, "all", "a", false, "Remove all unused data")
- 	flags.BoolVar(&pruneOptions.External, "external", false, "Remove container data in storage not controlled by podman")
-+	flags.BoolVar(&pruneOptions.Build, "build", false, "Remove build containers")
- 	flags.BoolVar(&pruneOptions.Volume, "volumes", false, "Prune volumes")
- 	filterFlagName := "filter"
- 	flags.StringArrayVar(&filters, filterFlagName, []string{}, "Provide filter values (e.g. 'label=<key>=<value>')")
-@@ -64,8 +65,12 @@ func prune(cmd *cobra.Command, args []string) error {
- 			volumeString = `
- 	- all volumes not used by at least one container`
- 		}
--
--		fmt.Printf(createPruneWarningMessage(pruneOptions), volumeString, "Are you sure you want to continue? [y/N] ")
-+		buildString := ""
-+		if pruneOptions.Build {
-+			buildString = `
-+	- all build containers`
-+		}
-+		fmt.Printf(createPruneWarningMessage(pruneOptions), volumeString, buildString, "Are you sure you want to continue? [y/N] ")
- 
- 		answer, err := reader.ReadString('\n')
- 		if err != nil {
-@@ -124,7 +129,7 @@ func createPruneWarningMessage(pruneOpts entities.SystemPruneOptions) string {
- 	if pruneOpts.All {
- 		return `WARNING! This command removes:
- 	- all stopped containers
--	- all networks not used by at least one container%s
-+	- all networks not used by at least one container%s%s
- 	- all images without at least one container associated with them
- 	- all build cache
- 
-@@ -132,7 +137,7 @@ func createPruneWarningMessage(pruneOpts entities.SystemPruneOptions) string {
- 	}
- 	return `WARNING! This command removes:
- 	- all stopped containers
--	- all networks not used by at least one container%s
-+	- all networks not used by at least one container%s%s
- 	- all dangling images
- 	- all dangling build cache
- 
-diff --git a/docs/source/markdown/podman-system-prune.1.md b/docs/source/markdown/podman-system-prune.1.md
-index 52f9ec1c7..95099d018 100644
---- a/docs/source/markdown/podman-system-prune.1.md
-+++ b/docs/source/markdown/podman-system-prune.1.md
-@@ -7,20 +7,28 @@ podman\-system\-prune - Remove all unused pods, containers, images, networks, an
- **podman system prune** [*options*]
- 
- ## DESCRIPTION
--**podman system prune** removes all unused containers (both dangling and unreferenced), pods, networks, and optionally, volumes from local storage.
-+**podman system prune** removes all unused containers (both dangling and unreferenced), build containers, pods, networks, and optionally, volumes from local storage.
- 
- Use the **--all** option to delete all unused images.  Unused images are dangling images as well as any image that does not have any containers based on it.
- 
- By default, volumes are not removed to prevent important data from being deleted if there is currently no container using the volume. Use the **--volumes** flag when running the command to prune volumes as well.
- 
-+By default, build containers are not removed to prevent interference with builds in progress. Use the **--build** flag when running the command to remove build containers as well.
-+
- ## OPTIONS
- #### **--all**, **-a**
- 
- Recursively remove all unused pods, containers, images, networks, and volume data. (Maximum 50 iterations.)
- 
-+#### **--build**
-+
-+Removes any build containers that were created during the build, but were not removed because the build was unexpectedly terminated.
-+
-+Note: **This is not safe operation and should be executed only when no builds are in progress. It can interfere with builds in progress.**
-+
- #### **--external**
- 
--Removes all leftover container storage files from local storage not managed by Podman. In normal circumstances, no such data exists, but in case of an unclean shutdown, the Podman database may be corrupted and cause this.
-+Tries to clean up remainders of previous containers or layers that are not references in the storage json files. These can happen in the case of unclean shutdowns or regular restarts in transient storage mode.
- 
- However, when using transient storage mode, the Podman database does not persist. This means containers leave the writable layers on disk after a reboot. When using a transient store, it is recommended that the **podman system prune --external** command is run during boot.
- 
-diff --git a/libpod/runtime.go b/libpod/runtime.go
-index 986e40f60..609fbba57 100644
---- a/libpod/runtime.go
-+++ b/libpod/runtime.go
-@@ -33,6 +33,7 @@ import (
- 	"github.com/containers/podman/v4/libpod/lock"
- 	"github.com/containers/podman/v4/libpod/plugin"
- 	"github.com/containers/podman/v4/libpod/shutdown"
-+	"github.com/containers/podman/v4/pkg/domain/entities/reports"
- 	"github.com/containers/podman/v4/pkg/rootless"
- 	"github.com/containers/podman/v4/pkg/systemd"
- 	"github.com/containers/podman/v4/pkg/util"
-@@ -1250,3 +1251,52 @@ func (r *Runtime) LockConflicts() (map[uint32][]string, []uint32, error) {
- 
- 	return toReturn, locksHeld, nil
- }
-+
-+// Exists checks whether a file or directory exists at the given path.
-+// If the path is a symlink, the symlink is followed.
-+func Exists(path string) error {
-+	// It uses unix.Faccessat which is a faster operation compared to os.Stat for
-+	// simply checking the existence of a file.
-+	err := unix.Faccessat(unix.AT_FDCWD, path, unix.F_OK, 0)
-+	if err != nil {
-+		return &os.PathError{Op: "faccessat", Path: path, Err: err}
-+	}
-+	return nil
-+}
-+
-+// PruneBuildContainers removes any build containers that were created during the build,
-+// but were not removed because the build was unexpectedly terminated.
-+//
-+// Note: This is not safe operation and should be executed only when no builds are in progress. It can interfere with builds in progress.
-+func (r *Runtime) PruneBuildContainers() ([]*reports.PruneReport, error) {
-+	stageContainersPruneReports := []*reports.PruneReport{}
-+
-+	containers, err := r.store.Containers()
-+	if err != nil {
-+		return stageContainersPruneReports, err
-+	}
-+	for _, container := range containers {
-+		path, err := r.store.ContainerDirectory(container.ID)
-+		if err != nil {
-+			return stageContainersPruneReports, err
-+		}
-+		if err := Exists(filepath.Join(path, "buildah.json")); err != nil {
-+			continue
-+		}
-+
-+		report := &reports.PruneReport{
-+			Id: container.ID,
-+		}
-+		size, err := r.store.ContainerSize(container.ID)
-+		if err != nil {
-+			report.Err = err
-+		}
-+		report.Size = uint64(size)
-+
-+		if err := r.store.DeleteContainer(container.ID); err != nil {
-+			report.Err = errors.Join(report.Err, err)
-+		}
-+		stageContainersPruneReports = append(stageContainersPruneReports, report)
-+	}
-+	return stageContainersPruneReports, nil
-+}
-diff --git a/pkg/api/handlers/libpod/system.go b/pkg/api/handlers/libpod/system.go
-index 70d4493f8..7c129b1ba 100644
---- a/pkg/api/handlers/libpod/system.go
-+++ b/pkg/api/handlers/libpod/system.go
-@@ -22,6 +22,7 @@ func SystemPrune(w http.ResponseWriter, r *http.Request) {
- 		All      bool `schema:"all"`
- 		Volumes  bool `schema:"volumes"`
- 		External bool `schema:"external"`
-+		Build    bool `schema:"build"`
- 	}{}
- 
- 	if err := decoder.Decode(&query, r.URL.Query()); err != nil {
-@@ -43,6 +44,7 @@ func SystemPrune(w http.ResponseWriter, r *http.Request) {
- 		Volume:   query.Volumes,
- 		Filters:  *filterMap,
- 		External: query.External,
-+		Build:    query.Build,
- 	}
- 	report, err := containerEngine.SystemPrune(r.Context(), pruneOptions)
- 	if err != nil {
-diff --git a/pkg/bindings/system/types.go b/pkg/bindings/system/types.go
-index 89e093f68..b4a4ff064 100644
---- a/pkg/bindings/system/types.go
-+++ b/pkg/bindings/system/types.go
-@@ -18,6 +18,7 @@ type PruneOptions struct {
- 	Filters  map[string][]string
- 	Volumes  *bool
- 	External *bool
-+	Build    *bool
- }
- 
- // VersionOptions are optional options for getting version info
-diff --git a/pkg/bindings/system/types_prune_options.go b/pkg/bindings/system/types_prune_options.go
-index d00498520..5f3bd652c 100644
---- a/pkg/bindings/system/types_prune_options.go
-+++ b/pkg/bindings/system/types_prune_options.go
-@@ -76,3 +76,18 @@ func (o *PruneOptions) GetExternal() bool {
- 	}
- 	return *o.External
- }
-+
-+// WithBuild set field Build to given value
-+func (o *PruneOptions) WithBuild(value bool) *PruneOptions {
-+	o.Build = &value
-+	return o
-+}
-+
-+// GetBuild returns value of field Build
-+func (o *PruneOptions) GetBuild() bool {
-+	if o.Build == nil {
-+		var z bool
-+		return z
-+	}
-+	return *o.Build
-+}
-diff --git a/pkg/domain/entities/system.go b/pkg/domain/entities/system.go
-index 473db3530..f6938652a 100644
---- a/pkg/domain/entities/system.go
-+++ b/pkg/domain/entities/system.go
-@@ -22,6 +22,7 @@ type SystemPruneOptions struct {
- 	Volume   bool
- 	Filters  map[string][]string `json:"filters" schema:"filters"`
- 	External bool
-+	Build    bool
- }
- 
- // SystemPruneReport provides report after system prune is executed.
-diff --git a/pkg/domain/infra/abi/system.go b/pkg/domain/infra/abi/system.go
-index 24ee64d29..ea3e5f203 100644
---- a/pkg/domain/infra/abi/system.go
-+++ b/pkg/domain/infra/abi/system.go
-@@ -150,16 +150,16 @@ func (ic *ContainerEngine) SetupRootless(_ context.Context, noMoveProcess bool)
- 	return nil
- }
- 
--// SystemPrune removes unused data from the system. Pruning pods, containers, networks, volumes and images.
-+// SystemPrune removes unused data from the system. Pruning pods, containers, build container, networks, volumes and images.
- func (ic *ContainerEngine) SystemPrune(ctx context.Context, options entities.SystemPruneOptions) (*entities.SystemPruneReport, error) {
- 	var systemPruneReport = new(entities.SystemPruneReport)
- 
- 	if options.External {
--		if options.All || options.Volume || len(options.Filters) > 0 {
-+		if options.All || options.Volume || len(options.Filters) > 0 || options.Build {
- 			return nil, fmt.Errorf("system prune --external cannot be combined with other options")
- 		}
--		err := ic.Libpod.GarbageCollect()
--		if err != nil {
-+
-+		if err := ic.Libpod.GarbageCollect(); err != nil {
- 			return nil, err
- 		}
- 		return systemPruneReport, nil
-@@ -170,6 +170,17 @@ func (ic *ContainerEngine) SystemPrune(ctx context.Context, options entities.Sys
- 		filters = append(filters, fmt.Sprintf("%s=%s", k, v[0]))
- 	}
- 	reclaimedSpace := (uint64)(0)
-+
-+	// Prune Build Containers
-+	if options.Build {
-+		stageContainersPruneReports, err := ic.Libpod.PruneBuildContainers()
-+		if err != nil {
-+			return nil, err
-+		}
-+		reclaimedSpace += reports.PruneReportsSize(stageContainersPruneReports)
-+		systemPruneReport.ContainerPruneReports = append(systemPruneReport.ContainerPruneReports, stageContainersPruneReports...)
-+	}
-+
- 	found := true
- 	for found {
- 		found = false
-diff --git a/pkg/domain/infra/tunnel/system.go b/pkg/domain/infra/tunnel/system.go
-index fc82e7b2b..142a9fa5c 100644
---- a/pkg/domain/infra/tunnel/system.go
-+++ b/pkg/domain/infra/tunnel/system.go
-@@ -19,7 +19,7 @@ func (ic *ContainerEngine) SetupRootless(_ context.Context, noMoveProcess bool)
- 
- // SystemPrune prunes unused data from the system.
- func (ic *ContainerEngine) SystemPrune(ctx context.Context, opts entities.SystemPruneOptions) (*entities.SystemPruneReport, error) {
--	options := new(system.PruneOptions).WithAll(opts.All).WithVolumes(opts.Volume).WithFilters(opts.Filters).WithExternal(opts.External)
-+	options := new(system.PruneOptions).WithAll(opts.All).WithVolumes(opts.Volume).WithFilters(opts.Filters).WithExternal(opts.External).WithBuild(opts.Build)
- 	return system.Prune(ic.ClientCtx, options)
- }
- 
-diff --git a/test/e2e/prune_test.go b/test/e2e/prune_test.go
-index 01e848478..57bd5582d 100644
---- a/test/e2e/prune_test.go
-+++ b/test/e2e/prune_test.go
-@@ -4,6 +4,8 @@ import (
- 	"fmt"
- 	"os"
- 	"path/filepath"
-+	"syscall"
-+	"time"
- 
- 	. "github.com/containers/podman/v4/test/utils"
- 	. "github.com/onsi/ginkgo/v2"
-@@ -22,6 +24,11 @@ FROM scratch
- ENV test1=test1
- ENV test2=test2`
- 
-+var longBuildImage = fmt.Sprintf(`
-+FROM %s
-+RUN echo "Hello, World!"
-+RUN RUN echo "Please use signal 9 this will never ends" && sleep 10000s`, ALPINE)
-+
- var _ = Describe("Podman prune", func() {
- 
- 	It("podman container prune containers", func() {
-@@ -593,4 +600,63 @@ var _ = Describe("Podman prune", func() {
- 		Expect(err).ToNot(HaveOccurred())
- 		Expect(dirents).To(HaveLen(3))
- 	})
-+
-+	It("podman system prune --build clean up after terminated build", func() {
-+		useCustomNetworkDir(podmanTest, tempdir)
-+
-+		podmanTest.BuildImage(pruneImage, "alpine_notleaker:latest", "false")
-+
-+		create := podmanTest.Podman([]string{"create", "--name", "test", BB, "sleep", "10000"})
-+		create.WaitWithDefaultTimeout()
-+		Expect(create).Should(ExitCleanly())
-+
-+		containerFilePath := filepath.Join(podmanTest.TempDir, "ContainerFile-podman-leaker")
-+		err := os.WriteFile(containerFilePath, []byte(longBuildImage), 0755)
-+		Expect(err).ToNot(HaveOccurred())
-+
-+		build := podmanTest.Podman([]string{"build", "-f", containerFilePath, "-t", "podmanleaker"})
-+		// Build will never finish so let's wait for build to ask for SIGKILL to simulate a failed build that leaves stage containers.
-+		matchedOutput := false
-+		for range 900 {
-+			if build.LineInOutputContains("Please use signal 9") {
-+				matchedOutput = true
-+				build.Signal(syscall.SIGKILL)
-+				break
-+			}
-+			time.Sleep(100 * time.Millisecond)
-+		}
-+		if !matchedOutput {
-+			Fail("Did not match special string in podman build")
-+		}
-+
-+		// Check Intermediate image of stage container
-+		none := podmanTest.Podman([]string{"images", "-a"})
-+		none.WaitWithDefaultTimeout()
-+		Expect(none).Should(ExitCleanly())
-+		Expect(none.OutputToString()).Should(ContainSubstring("none"))
-+
-+		// Check if Container and Stage Container exist
-+		count := podmanTest.Podman([]string{"ps", "-aq", "--external"})
-+		count.WaitWithDefaultTimeout()
-+		Expect(count).Should(ExitCleanly())
-+		Expect(count.OutputToStringArray()).To(HaveLen(3))
-+
-+		prune := podmanTest.Podman([]string{"system", "prune", "--build", "-f"})
-+		prune.WaitWithDefaultTimeout()
-+		Expect(prune).Should(ExitCleanly())
-+
-+		// Container should still exist, but no stage containers
-+		count = podmanTest.Podman([]string{"ps", "-aq", "--external"})
-+		count.WaitWithDefaultTimeout()
-+		Expect(count).Should(ExitCleanly())
-+		Expect(count.OutputToString()).To(BeEmpty())
-+
-+		Expect(podmanTest.NumberOfContainers()).To(Equal(0))
-+
-+		after := podmanTest.Podman([]string{"images", "-a"})
-+		after.WaitWithDefaultTimeout()
-+		Expect(after).Should(ExitCleanly())
-+		Expect(after.OutputToString()).ShouldNot(ContainSubstring("none"))
-+		Expect(after.OutputToString()).Should(ContainSubstring("notleaker"))
-+	})
- })
-

.github/scripts/s390x-ci/self-hosted-builder/podman-patches/podman-25245.patch

@@ -1,21 +0,0 @@
-diff --git a/pkg/rootless/rootless_linux.c b/pkg/rootless/rootless_linux.c
-index 4f71d49e5c..3d74af6a6c 100644
---- a/pkg/rootless/rootless_linux.c
-+++ b/pkg/rootless/rootless_linux.c
-@@ -658,7 +658,7 @@ create_pause_process (const char *pause_pid_file_path, char **argv)
-   if (pipe (p) < 0)
-     return -1;
-
--  pid = fork ();
-+  pid = syscall_clone (SIGCHLD, NULL);
-   if (pid < 0)
-     {
-       close (p[0]);
-@@ -689,7 +689,7 @@ create_pause_process (const char *pause_pid_file_path, char **argv)
-       close (p[0]);
-
-       setsid ();
--      pid = fork ();
-+      pid = syscall_clone (SIGCHLD, NULL);
-       if (pid < 0)
-         _exit (EXIT_FAILURE);

.github/scripts/s390x-ci/tests_list.py

@@ -1,97 +0,0 @@
-#!/usr/bin/env python3
-
-import os
-import re
-import sys
-
-
-sys.path.insert(1, os.path.join(sys.path[0], "..", "..", ".."))
-
-from tools.testing.discover_tests import TESTS
-
-
-skip_list = [
-    # these tests fail due to various reasons
-    "dynamo/test_misc",
-    "inductor/test_aot_inductor",
-    "inductor/test_cpu_repro",
-    "inductor/test_cpu_select_algorithm",
-    "inductor/test_aot_inductor_arrayref",
-    "inductor/test_torchinductor_codegen_dynamic_shapes",
-    "lazy/test_meta_kernel",
-    "onnx/test_utility_funs",
-    "profiler/test_profiler",
-    "test_ao_sparsity",
-    "test_cpp_extensions_open_device_registration",
-    "test_jit",
-    "test_metal",
-    "test_mps",
-    "dynamo/test_torchrec",
-    "inductor/test_aot_inductor_utils",
-    "inductor/test_coordinate_descent_tuner",
-    "test_jiterator",
-    # these tests run long and fail in addition to that
-    "dynamo/test_dynamic_shapes",
-    "test_quantization",
-    "inductor/test_torchinductor",
-    "inductor/test_torchinductor_dynamic_shapes",
-    "inductor/test_torchinductor_opinfo",
-    "test_binary_ufuncs",
-    "test_unary_ufuncs",
-    # these tests fail when cuda is not available
-    "inductor/test_cudacodecache",
-    "inductor/test_inductor_utils",
-    "inductor/test_inplacing_pass",
-    "inductor/test_kernel_benchmark",
-    "inductor/test_max_autotune",
-    "inductor/test_move_constructors_to_cuda",
-    "inductor/test_multi_kernel",
-    "inductor/test_pattern_matcher",
-    "inductor/test_perf",
-    "inductor/test_select_algorithm",
-    "inductor/test_snode_runtime",
-    "inductor/test_triton_wrapper",
-    # these tests fail when mkldnn is not available
-    "inductor/test_custom_post_grad_passes",
-    "inductor/test_mkldnn_pattern_matcher",
-    # lacks quantization support
-    "onnx/test_models_quantized_onnxruntime",
-    "onnx/test_pytorch_onnx_onnxruntime",
-    # https://github.com/pytorch/pytorch/issues/102078
-    "test_decomp",
-    # https://github.com/pytorch/pytorch/issues/146698
-    "test_model_exports_to_core_aten",
-    # runs very long, skip for now
-    "inductor/test_layout_optim",
-    "test_fx",
-    # some false errors
-    "doctests",
-]
-
-skip_list_regex = [
-    # distributed tests fail randomly
-    "distributed/.*",
-]
-
-all_testfiles = sorted(TESTS)
-
-filtered_testfiles = []
-
-for filename in all_testfiles:
-    if filename in skip_list:
-        continue
-
-    regex_filtered = False
-
-    for regex_string in skip_list_regex:
-        if re.fullmatch(regex_string, filename):
-            regex_filtered = True
-            break
-
-    if regex_filtered:
-        continue
-
-    filtered_testfiles.append(filename)
-
-for filename in filtered_testfiles:
-    print('    "' + filename + '",')

.github/scripts/trymerge.py

@@ -819,9 +819,10 @@ class GitHubPR:
                     cursor=info["reviews"]["pageInfo"]["startCursor"],
                 )
                 info = rc["data"]["repository"]["pullRequest"]
-        reviews = {
-            author: state for author, state in self._reviews if state != "COMMENTED"
-        }
+        reviews = {}
+        for author, state in self._reviews:
+            if state != "COMMENTED":
+                reviews[author] = state
         return list(reviews.items())
 
     def get_approved_by(self) -> list[str]:
@@ -2281,8 +2282,7 @@ def merge(
         except MandatoryChecksMissingError as ex:
             last_exception = str(ex)
             print(
-                f"Merge of https://github.com/{pr.org}/{pr.project}/pull/{pr.pr_num} failed due to: {ex}. Retrying in 5 min",
-                flush=True,
+                f"Merge of https://github.com/{pr.org}/{pr.project}/pull/{pr.pr_num} failed due to: {ex}. Retrying in 5 min"
             )
             time.sleep(5 * 60)
     # Finally report timeout back

.github/scripts/windows/build_triton.bat

@@ -9,7 +9,8 @@ if "%PY_VERS%" == "3.13t" (
 ) else (
     call conda create -n %PYTHON_PREFIX% -y -c=conda-forge python=%PY_VERS%
 )
-call conda run -n %PYTHON_PREFIX% pip install wheel pybind11 certifi cython cmake setuptools==72.1.0 ninja
+:: Fix cmake version for issue https://github.com/pytorch/pytorch/issues/150480
+call conda run -n %PYTHON_PREFIX% pip install wheel pybind11 certifi cython cmake==3.31.6 setuptools==72.1.0 ninja
 
 dir "%VC_INSTALL_PATH%"
 

.github/templates/common.yml.j2

@@ -32,7 +32,7 @@ concurrency:
 {%- macro setup_ec2_windows() -%}
       !{{ display_ec2_information() }}
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}

.github/templates/linux_binary_build_workflow.yml.j2

@@ -53,7 +53,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -147,9 +147,9 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ runner.temp }}/artifacts/"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: !{{ config["container_image"] }}
       - name: Test Pytorch binary
@@ -168,12 +168,12 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ runner.temp }}/artifacts/"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
       - name: ROCm set GPU_FLAG
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: !{{ config["container_image"] }}
       - name: Test Pytorch binary

.github/templates/macos_binary_build_workflow.yml.j2

@@ -76,7 +76,7 @@ jobs:
           elif [ -d "/Applications/Xcode_13.3.1.app" ]; then
             echo "DEVELOPER_DIR=/Applications/Xcode_13.3.1.app/Contents/Developer" >> "${GITHUB_ENV}"
           fi
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
       - name: Populate binary env
         run: |
           # shellcheck disable=SC1091

.github/templates/upload.yml.j2

@@ -25,6 +25,9 @@
       DOCKER_IMAGE: !{{ config["container_image"] }}
 {%- endif %}
 {%- if config["package_type"] == "manywheel" %}
+  {%- if config["devtoolset"] %}
+      DESIRED_DEVTOOLSET: !{{ config["devtoolset"] }}
+  {%- endif %}
   {%- if config.use_split_build is defined %}
       use_split_build: !{{ config["use_split_build"] }}
   {%- endif %}
@@ -34,6 +37,9 @@
       LIBTORCH_CONFIG: !{{ config["libtorch_config"] }}
   {%- endif %}
       LIBTORCH_VARIANT: !{{ config["libtorch_variant"] }}
+  {%- if config["devtoolset"] %}
+      DESIRED_DEVTOOLSET: !{{ config["devtoolset"] }}
+  {%- endif %}
   {%- if is_windows %}
       # This is a dummy value for libtorch to work correctly with our batch scripts
       # without this value pip does not get installed for some reason

.github/templates/windows_arm64_binary_build_workflow.yml.j2

@@ -55,7 +55,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/templates/windows_binary_build_workflow.yml.j2

@@ -55,7 +55,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -79,7 +79,7 @@ jobs:
     steps:
       !{{ common.setup_ec2_windows() }}
       !{{ set_runner_specific_vars() }}
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
       - name: Populate binary env
         shell: bash
         run: |
@@ -124,7 +124,7 @@ jobs:
         with:
           name: !{{ config["build_name"] }}
           path: "${{ env.PYTORCH_FINAL_PACKAGE_DIR }}"
-      !{{ common.checkout(deep_clone=False, directory="pytorch") }}
+      !{{ common.checkout(deep_clone=False, directory="pytorch", checkout_pr_head=False) }}
       - name: Populate binary env
         shell: bash
         run: |

.github/workflows/_bazel-build-test.yml

@@ -47,7 +47,7 @@ jobs:
       reenabled-issues: ${{ steps.filter.outputs.reenabled-issues }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           fetch-depth: 1
           submodules: false
@@ -69,25 +69,25 @@ jobs:
     runs-on: ${{ matrix.runner }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -97,7 +97,7 @@ jobs:
         run: echo "IN_CONTAINER_RUNNER=$(if [ -f /.inarc ] || [ -f /.incontainer ]; then echo true ; else echo false; fi)" >> "$GITHUB_OUTPUT"
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.7
         if: ${{ inputs.cuda-version != 'cpu' && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' }}
 
       - name: Output disk space left
@@ -209,5 +209,5 @@ jobs:
           file-suffix: bazel-${{ github.job }}_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always()

.github/workflows/_binary-build-linux.yml

@@ -70,6 +70,10 @@ on:
         required: false
         type: string
         description: Desired libtorch variant (for libtorch builds only)
+      DESIRED_DEVTOOLSET:
+        required: false
+        type: string
+        description: Desired dev toolset
       DESIRED_PYTHON:
         required: false
         type: string
@@ -100,6 +104,7 @@ jobs:
       SKIP_ALL_TESTS: 1
       LIBTORCH_CONFIG: ${{ inputs.LIBTORCH_CONFIG }}
       LIBTORCH_VARIANT: ${{ inputs.LIBTORCH_VARIANT }}
+      DESIRED_DEVTOOLSET: ${{ inputs.DESIRED_DEVTOOLSET }}
       DESIRED_PYTHON: ${{ inputs.DESIRED_PYTHON }}
       PYTORCH_EXTRA_INSTALL_REQUIREMENTS: ${{ inputs.PYTORCH_EXTRA_INSTALL_REQUIREMENTS }}
       ALPINE_IMAGE: ${{ inputs.ALPINE_IMAGE }}
@@ -125,6 +130,7 @@ jobs:
             echo "SKIP_ALL_TESTS=${{ env.SKIP_ALL_TESTS }}"
             echo "LIBTORCH_CONFIG=${{ env.LIBTORCH_CONFIG }}"
             echo "LIBTORCH_VARIANT=${{ env.LIBTORCH_VARIANT }}"
+            echo "DESIRED_DEVTOOLSET=${{ env.DESIRED_DEVTOOLSET }}"
             echo "DESIRED_PYTHON=${{ env.DESIRED_PYTHON }}"
             echo "PYTORCH_EXTRA_INSTALL_REQUIREMENTS=${{ env.PYTORCH_EXTRA_INSTALL_REQUIREMENTS }}"
             echo "ALPINE_IMAGE=${{ env.ALPINE_IMAGE }}"
@@ -144,13 +150,13 @@ jobs:
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
         if: inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         continue-on-error: true
         with:
           github-secret: ${{ secrets.github-token }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' || inputs.build_environment == 'linux-s390x-binary-manywheel' }}
 
@@ -180,7 +186,6 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: actions/checkout@v4
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -205,7 +210,7 @@ jobs:
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ inputs.DOCKER_IMAGE }}
 
@@ -218,6 +223,7 @@ jobs:
             -e BINARY_ENV_FILE \
             -e BUILD_ENVIRONMENT \
             -e DESIRED_CUDA \
+            -e DESIRED_DEVTOOLSET \
             -e DESIRED_PYTHON \
             -e GITHUB_ACTIONS \
             -e GPU_ARCH_TYPE \
@@ -260,7 +266,7 @@ jobs:
 
       - name: Teardown Linux
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
 
       - name: Chown workspace
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'

.github/workflows/_binary-test-linux.yml

@@ -47,6 +47,10 @@ on:
         required: false
         type: string
         description: Desired libtorch variant (for libtorch builds only)
+      DESIRED_DEVTOOLSET:
+        required: false
+        type: string
+        description: Desired dev toolset
       DESIRED_PYTHON:
         required: false
         type: string
@@ -88,6 +92,7 @@ jobs:
       SKIP_ALL_TESTS: 1
       LIBTORCH_CONFIG: ${{ inputs.LIBTORCH_CONFIG }}
       LIBTORCH_VARIANT: ${{ inputs.LIBTORCH_VARIANT }}
+      DESIRED_DEVTOOLSET: ${{ inputs.DESIRED_DEVTOOLSET }}
       DESIRED_PYTHON: ${{ inputs.DESIRED_PYTHON }}
       ALPINE_IMAGE: ${{ inputs.ALPINE_IMAGE }}
       AWS_DEFAULT_REGION: us-east-1
@@ -113,6 +118,7 @@ jobs:
             echo "SKIP_ALL_TESTS=${{ env.SKIP_ALL_TESTS }}"
             echo "LIBTORCH_CONFIG=${{ env.LIBTORCH_CONFIG }}"
             echo "LIBTORCH_VARIANT=${{ env.LIBTORCH_VARIANT }}"
+            echo "DESIRED_DEVTOOLSET=${{ env.DESIRED_DEVTOOLSET }}"
             echo "DESIRED_PYTHON=${{ env.DESIRED_PYTHON }}"
 
             echo "ALPINE_IMAGE=${{ env.ALPINE_IMAGE }}"
@@ -127,14 +133,14 @@ jobs:
 
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
         if: inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         continue-on-error: true
         with:
           github-secret: ${{ secrets.github-token }}
 
         # Setup the environment
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: ${{ inputs.build_environment == 'linux-aarch64-binary-manywheel' || inputs.build_environment == 'linux-s390x-binary-manywheel' }}
 
@@ -157,7 +163,6 @@ jobs:
       - name: Checkout PyTorch to pytorch dir
         uses: actions/checkout@v4
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           show-progress: false
           path: pytorch
@@ -188,12 +193,12 @@ jobs:
           path: "${{ runner.temp }}/artifacts/"
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.7
         if: ${{ inputs.GPU_ARCH_TYPE == 'cuda' && steps.filter.outputs.is-test-matrix-empty == 'False' }}
 
       - name: Pull Docker image
         if: ${{ steps.filter.outputs.is-test-matrix-empty == 'False' && inputs.build_environment != 'linux-s390x-binary-manywheel' }}
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ inputs.DOCKER_IMAGE }}
 
@@ -203,7 +208,7 @@ jobs:
 
       - name: Teardown Linux
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
 
       - name: Chown workspace
         if: always() && inputs.build_environment != 'linux-s390x-binary-manywheel'

.github/workflows/_binary-upload.yml

@@ -43,6 +43,10 @@ on:
         required: false
         type: string
         description: Desired libtorch variant (for libtorch builds only)
+      DESIRED_DEVTOOLSET:
+        required: false
+        type: string
+        description: Desired dev toolset
       DESIRED_PYTHON:
         required: false
         type: string
@@ -76,6 +80,7 @@ jobs:
       SKIP_ALL_TESTS: 1
       LIBTORCH_CONFIG: ${{ inputs.LIBTORCH_CONFIG }}
       LIBTORCH_VARIANT: ${{ inputs.LIBTORCH_VARIANT }}
+      DESIRED_DEVTOOLSET: ${{ inputs.DESIRED_DEVTOOLSET }}
       DESIRED_PYTHON: ${{ inputs.DESIRED_PYTHON }}
       BINARY_ENV_FILE: /tmp/env
       GITHUB_TOKEN: ${{ secrets.github-token }}
@@ -85,7 +90,7 @@ jobs:
       USE_SPLIT_BUILD: ${{ inputs.use_split_build }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 

.github/workflows/_docs.yml

@@ -84,7 +84,7 @@ jobs:
     name: build-docs-${{ matrix.docs_type }}-${{ inputs.push }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -95,7 +95,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
@@ -110,12 +110,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -222,5 +222,5 @@ jobs:
           s3-prefix: pytorch/pytorch/${{ github.event.pull_request.number }}/functorchdocs
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always()

.github/workflows/_linux-build.yml

@@ -106,7 +106,7 @@ jobs:
       test-matrix: ${{ steps.filter.outputs.test-matrix }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -116,7 +116,7 @@ jobs:
       # checkout because when we run this action we don't *have* a local
       # checkout. In other cases you should prefer a local checkout.
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 
@@ -134,7 +134,7 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image-name: ${{ inputs.docker-image-name }}
@@ -150,7 +150,7 @@ jobs:
           echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
@@ -312,7 +312,7 @@ jobs:
           build-time: ${{ steps.build.outputs.build_time }}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always() && inputs.build-environment != 'linux-s390x-binary-manywheel'
 
       - name: Cleanup docker

.github/workflows/_linux-test.yml

@@ -80,7 +80,7 @@ jobs:
     timeout-minutes: ${{ matrix.mem_leak_check == 'mem_leak_check' && 600 || inputs.timeout-minutes }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         if: ${{ !contains(matrix.runner, 'gcp.a100') && inputs.build-environment != 'linux-s390x-binary-manywheel' }}
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -89,7 +89,7 @@ jobs:
               docker exec -it $(docker container ps --format '{{.ID}}') bash
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 
@@ -107,7 +107,7 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image-name: ${{ inputs.docker-image }}
@@ -123,7 +123,7 @@ jobs:
           echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         if: inputs.build-environment != 'linux-s390x-binary-manywheel'
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
@@ -135,7 +135,7 @@ jobs:
 
       - name: Install nvidia driver, nvidia-docker runtime, set GPU_FLAG
         id: install-nvidia-driver
-        uses: pytorch/test-infra/.github/actions/setup-nvidia@main
+        uses: pytorch/test-infra/.github/actions/setup-nvidia@release/2.7
         if: ${{ contains(inputs.build-environment, 'cuda') && !contains(matrix.config, 'nogpu') && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false' }}
 
       - name: Setup GPU_FLAG for docker run
@@ -371,7 +371,7 @@ jobs:
           job_identifier: ${{ github.workflow }}_${{ inputs.build-environment }}
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.7
         with:
           benchmark-results-dir: test/test-reports
           dry-run: false
@@ -428,7 +428,7 @@ jobs:
           workflow_attempt: ${{github.run_attempt}}
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always() && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false'
 
       # NB: We are currently having an intermittent GPU-related issue on G5 runners with

.github/workflows/_mac-build.yml

@@ -33,6 +33,10 @@ on:
         default: "3.9"
         description: |
           The python version to be used. Will be 3.9 by default
+      environment-file:
+        required: false
+        type: string
+        description: Set the conda environment file used to setup macOS build.
       test-matrix:
         required: false
         type: string
@@ -67,11 +71,11 @@ jobs:
       test-matrix: ${{ steps.filter.outputs.test-matrix }}
     steps:
       - name: Clean up disk space before running MacOS workflow
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.7
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Set xcode version
         env:
@@ -82,12 +86,23 @@ jobs:
           fi
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        if: inputs.environment-file == ''
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.7
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}
           pip-requirements-file: .github/requirements/pip-requirements-${{ runner.os }}.txt
 
+      # This option is used when cross-compiling arm64 from x86-64. Specifically, we need arm64 conda
+      # environment even though the arch is x86-64
+      - name: Setup miniconda using the provided environment file
+        if: inputs.environment-file != ''
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.7
+        with:
+          python-version: ${{ inputs.python-version }}
+          environment-file: ${{ inputs.environment-file }}
+          pip-requirements-file: .github/requirements/pip-requirements-${{ runner.os }}.txt
+
       - name: Install sccache (only for non-forked PRs, and pushes to trunk)
         uses: nick-fields/retry@v3.0.0
         if: ${{ github.event_name == 'push' || github.event.pull_request.head.repo.full_name == github.repository }}
@@ -192,4 +207,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.7

.github/workflows/_mac-test-mps.yml

@@ -41,7 +41,7 @@ jobs:
       reenabled-issues: ${{ steps.filter.outputs.reenabled-issues }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
 
@@ -82,7 +82,7 @@ jobs:
           use-gha: true
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.7
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}
@@ -170,4 +170,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.7

.github/workflows/_mac-test.yml

@@ -82,11 +82,11 @@ jobs:
           done
 
       - name: Clean up disk space before running MacOS workflow
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.7
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Start monitoring script
         id: monitor-script
@@ -109,7 +109,7 @@ jobs:
           use-gha: true
 
       - name: Setup miniconda
-        uses: pytorch/test-infra/.github/actions/setup-miniconda@main
+        uses: pytorch/test-infra/.github/actions/setup-miniconda@release/2.7
         with:
           python-version: ${{ inputs.python-version }}
           environment-file: .github/requirements/conda-env-${{ runner.os }}-${{ runner.arch }}
@@ -224,7 +224,7 @@ jobs:
           file-suffix: ${{ github.job }}-${{ matrix.config }}-${{ matrix.shard }}-${{ matrix.num_shards }}-${{ matrix.runner }}_${{ steps.get-job-id.outputs.job-id }}
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.7
         with:
           benchmark-results-dir: test/test-reports
           dry-run: false
@@ -234,4 +234,4 @@ jobs:
       - name: Clean up disk space
         if: always()
         continue-on-error: true
-        uses: pytorch/test-infra/.github/actions/check-disk-space@main
+        uses: pytorch/test-infra/.github/actions/check-disk-space@release/2.7

.github/workflows/_rocm-test.yml

@@ -70,7 +70,7 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 
@@ -92,12 +92,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 
@@ -302,7 +302,7 @@ jobs:
           aws-region: us-east-1
 
       - name: Upload the benchmark results
-        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@main
+        uses: pytorch/test-infra/.github/actions/upload-benchmark-results@release/2.7
         with:
           benchmark-results-dir: test/test-reports
           dry-run: false

.github/workflows/_runner-determinator.yml

@@ -54,7 +54,7 @@ jobs:
       PR_NUMBER: ${{ github.event.pull_request.number }}
     steps:
       # - name: Checkout PyTorch
-      #   uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+      #   uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
       #   with:
       #     fetch-depth: 1
       #     submodules: true

.github/workflows/_win-build.yml

@@ -84,10 +84,10 @@ jobs:
           git config --global core.fsmonitor false
 
       - name: Clean up leftover processes on non-ephemeral Windows runner
-        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@release/2.7
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -102,7 +102,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 

.github/workflows/_win-test.yml

@@ -66,10 +66,10 @@ jobs:
           git config --global core.fsmonitor false
 
       - name: Clean up leftover processes on non-ephemeral Windows runner
-        uses: pytorch/test-infra/.github/actions/cleanup-runner@main
+        uses: pytorch/test-infra/.github/actions/cleanup-runner@release/2.7
 
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
           instructions: |
@@ -85,7 +85,7 @@ jobs:
 
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           no-sudo: true
 

.github/workflows/_xpu-test.yml

@@ -62,7 +62,7 @@ jobs:
     steps:
       # [see note: pytorch repo ref]
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Setup XPU
         uses: ./.github/actions/setup-xpu
@@ -80,12 +80,12 @@ jobs:
 
       - name: Calculate docker image
         id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
           docker-image-name: ${{ inputs.docker-image }}
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
 

.github/workflows/build-almalinux-images.yml

@@ -41,12 +41,12 @@ jobs:
       CUDA_VERSION: ${{ matrix.cuda_version }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: almalinux-builder${{ matrix.cuda_version == 'cpu' && '-' || '-cuda' }}${{matrix.cuda_version}}
             docker-build-dir:  .ci/docker/almalinux

.github/workflows/build-libtorch-images.yml

@@ -32,7 +32,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -51,12 +51,12 @@ jobs:
       GPU_ARCH_VERSION: ${{ matrix.cuda_version }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: libtorch-cxx11-builder-cuda${{matrix.cuda_version}}
             docker-build-dir:  .ci/docker/libtorch
@@ -93,12 +93,12 @@ jobs:
       GPU_ARCH_VERSION: ${{ matrix.rocm_version }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: libtorch-cxx11-builder-rocm${{matrix.rocm_version}}
             docker-build-dir:  .ci/docker/libtorch
@@ -129,12 +129,12 @@ jobs:
     runs-on: "${{ needs.get-label-type.outputs.label-type }}linux.9xlarge.ephemeral"
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: libtorch-cxx11-builder-cpu
             docker-build-dir:  .ci/docker/libtorch

.github/workflows/build-manywheel-images-s390x.yml

@@ -41,7 +41,7 @@ jobs:
       GPU_ARCH_TYPE: cpu-s390x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
           no-sudo: true
@@ -62,12 +62,7 @@ jobs:
         if: cancelled()
         shell: bash
         run: |
-          # If podman build command is interrupted,
+          # if podman build command is interrupted,
           # it can leave a couple of processes still running.
-          # Order them to stop for clean shutdown.
-          # It looks like sometimes some processes remain
-          # after first cleanup.
-          # Wait a bit and do cleanup again. It looks like it helps.
-          docker system prune --build -f || true
-          sleep 60
+          # order them to stop for clean shutdown.
           docker system prune --build -f || true

.github/workflows/build-manywheel-images.yml

@@ -36,7 +36,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -57,12 +57,12 @@ jobs:
       - name: Purge tools folder (free space for build)
         run: rm -rf /opt/hostedtoolcache
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinux2_28-builder-cuda${{matrix.cuda_version}}
             docker-build-dir:  .ci/docker/manywheel
@@ -102,7 +102,7 @@ jobs:
         uses: actions/checkout@v3
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinuxaarch64-builder-cuda${{matrix.cuda_version}}
             docker-build-dir:  .ci/docker/manywheel
@@ -139,12 +139,12 @@ jobs:
       GPU_ARCH_VERSION: ${{ matrix.rocm_version }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinux2_28-builder-rocm${{matrix.rocm_version}}
             docker-build-dir:  .ci/docker/manywheel
@@ -177,12 +177,12 @@ jobs:
       GPU_ARCH_TYPE: cpu-manylinux_2_28
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinux2_28-builder-cpu
             docker-build-dir:  .ci/docker/manywheel
@@ -215,12 +215,12 @@ jobs:
       GPU_ARCH_TYPE: cpu-aarch64
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinuxaarch64-builder-cpu-aarch64
             docker-build-dir:  .ci/docker/manywheel
@@ -253,12 +253,12 @@ jobs:
       GPU_ARCH_TYPE: cpu-aarch64-2_28
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinux2_28_aarch64-builder-cpu-aarch64
             docker-build-dir:  .ci/docker/manywheel
@@ -294,12 +294,12 @@ jobs:
       GPU_ARCH_TYPE: cpu-cxx11-abi
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinuxcxx11-abi-builder-cpu-cxx11-abi
             docker-build-dir:  .ci/docker/manywheel
@@ -332,12 +332,12 @@ jobs:
       GPU_ARCH_TYPE: xpu
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
       - name: Calculate docker image
         if: env.WITH_PUSH == 'false'
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
             docker-image-name: manylinux2_28-builder-xpu
             docker-build-dir:  .ci/docker/manywheel

.github/workflows/build-triton-wheel.yml

@@ -3,7 +3,7 @@ name: Build Triton wheels
 on:
   push:
     branches:
-      - main
+      - release/2.7
     tags:
       # NOTE: Binary build pipelines should only get triggered on release candidate builds
       # Release candidate tags look like: v1.11.0-rc1
@@ -12,6 +12,8 @@ on:
       - .github/workflows/build-triton-wheel.yml
       - .github/scripts/build_triton_wheel.py
       - .github/ci_commit_pins/triton.txt
+      - .github/scripts/windows/install_vs2022.ps1
+      - .github/scripts/windows/build_triton.bat
       - .ci/docker/ci_commit_pins/triton.txt
       - .ci/docker/ci_commit_pins/triton-xpu.txt
   pull_request:
@@ -19,6 +21,8 @@ on:
       - .github/workflows/build-triton-wheel.yml
       - .github/scripts/build_triton_wheel.py
       - .github/ci_commit_pins/triton.txt
+      - .github/scripts/windows/install_vs2022.ps1
+      - .github/scripts/windows/build_triton.bat
       - .ci/docker/ci_commit_pins/triton.txt
       - .ci/docker/ci_commit_pins/triton-xpu.txt
 
@@ -30,7 +34,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -68,12 +72,12 @@ jobs:
       PLATFORM: 'manylinux_2_28_x86_64'
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
 
@@ -81,7 +85,7 @@ jobs:
         uses: ./.github/actions/setup-linux
 
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ env.DOCKER_IMAGE }}
 
@@ -133,7 +137,7 @@ jobs:
           fi
 
           docker exec -t "${container_name}" yum install -y zlib-devel zip
-          docker exec -t "${container_name}" "${PYTHON_EXECUTABLE}"  -m pip install -U setuptools==67.4.0 pybind11==2.13.1 auditwheel wheel
+          docker exec -t "${container_name}" "${PYTHON_EXECUTABLE}"  -m pip install -U setuptools==78.1.0 pybind11==2.13.1 auditwheel wheel
 
           if [[ ("${{ matrix.device }}" == "cuda" || "${{ matrix.device }}" == "rocm" || "${{ matrix.device }}" == "aarch64" ) ]]; then
             # With this install, it gets clang 16.0.6.
@@ -163,7 +167,7 @@ jobs:
           path: ${{ runner.temp }}/artifacts/wheelhouse/*
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always()
 
   build-wheel-win:
@@ -196,7 +200,7 @@ jobs:
           echo "instance-type: $(get_ec2_metadata instance-type)"
           echo "system info $(uname -a)"
       - name: "[FB EMPLOYEES] Enable SSH (Click me for login details)"
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         continue-on-error: true
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
@@ -243,7 +247,6 @@ jobs:
           .github/scripts/windows/build_triton.bat
           mkdir -p "${RUNNER_TEMP}/artifacts/"
           mv ./*.whl "${RUNNER_TEMP}/artifacts/"
-
       - uses: actions/upload-artifact@v4.4.0
         with:
           name: pytorch-triton-wheel-${{ matrix.py_vers }}-${{ matrix.device }}

.github/workflows/check-labels.yml

@@ -38,7 +38,7 @@ jobs:
     runs-on: linux.24_04.4x
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
           fetch-depth: 1

.github/workflows/close-nonexistent-disable-issues.yml

@@ -13,7 +13,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           submodules: false
           fetch-depth: 1

.github/workflows/create_release.yml

@@ -19,7 +19,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}

.github/workflows/docker-builds.yml

@@ -33,7 +33,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -99,21 +99,21 @@ jobs:
       # [see note: pytorch repo ref]
       # deep clone (fetch-depth 0) required for git merge-base
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
 
       - name: Setup Linux
         uses: ./.github/actions/setup-linux
 
       - name: Build docker image
         id: build-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@release/2.7
         with:
           docker-image-name: ${{ matrix.docker-image-name }}
           always-rebuild: true
           push: true
 
       - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
           docker-image: ${{ steps.build-docker-image.outputs.docker-image }}
 
@@ -145,5 +145,5 @@ jobs:
         if: always()
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always()

.github/workflows/docker-cache-mi300.yml

@@ -1,55 +0,0 @@
-name: docker-cache-mi300
-
-on:
-  # run every 6 hours
-  schedule:
-    - cron: 0 0,6,12,18 * * *
-  workflow_dispatch:
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.sha }}-${{ github.event_name }}
-  cancel-in-progress: true
-
-permissions:
-  id-token: write
-  contents: read
-
-jobs:
-  docker-cache:
-    if: github.repository_owner == 'pytorch'
-    runs-on: rocm-docker
-    steps:
-      - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
-        with:
-          no-sudo: true
-
-      - name: configure aws credentials
-        id: aws_creds
-        uses: aws-actions/configure-aws-credentials@v4
-        with:
-          role-to-assume: arn:aws:iam::308535385114:role/gha_workflow_s3_and_ecr_read_only
-          aws-region: us-east-1
-          role-duration-seconds: 18000
-
-      - name: Login to Amazon ECR
-        id: login-ecr
-        continue-on-error: false
-        uses: aws-actions/amazon-ecr-login@v2
-
-      - name: Calculate docker image
-        id: calculate-docker-image
-        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
-        with:
-          docker-image-name: pytorch-linux-focal-rocm-n-py3
-          push: false
-
-      - name: Pull docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
-        with:
-          docker-image: ${{ steps.calculate-docker-image.outputs.docker-image }}
-
-      - name: Tar and upload to S3 bucket
-        run: |
-          sudo docker save -o ~/docker-data/pytorch/pytorch_docker_image.tar ${{ steps.calculate-docker-image.outputs.docker-image }}
-          sudo rclone copy -P --s3-upload-concurrency 64 --s3-chunk-size 200M --s3-upload-cutoff 300M ~/docker-data/pytorch/pytorch_docker_image.tar oci:pytorchbucket0002/pytorch_docker_image --progress

.github/workflows/docker-release.yml

@@ -37,7 +37,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -52,7 +52,7 @@ jobs:
       matrix: ${{ steps.generate-matrix.outputs.matrix }}
     steps:
       - name: Checkout PyTorch
-        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@release/2.7
         with:
           fetch-depth: 1
           submodules: true
@@ -82,7 +82,7 @@ jobs:
       CUDNN_VERSION: ${{ matrix.cudnn_version }}
     steps:
       - name: Setup SSH (Click me for login details)
-        uses: pytorch/test-infra/.github/actions/setup-ssh@main
+        uses: pytorch/test-infra/.github/actions/setup-ssh@release/2.7
         with:
           github-secret: ${{ secrets.GITHUB_TOKEN }}
       # [see note: pytorch repo ref]
@@ -164,12 +164,12 @@ jobs:
           fi
 
       - name: Teardown Linux
-        uses: pytorch/test-infra/.github/actions/teardown-linux@main
+        uses: pytorch/test-infra/.github/actions/teardown-linux@release/2.7
         if: always()
 
   validate:
     needs: build
-    uses: pytorch/test-infra/.github/workflows/validate-docker-images.yml@main
+    uses: pytorch/test-infra/.github/workflows/validate-docker-images.yml@release/2.7
     with:
-      channel: nightly
+      channel: test
       ref: main

.github/workflows/generated-linux-aarch64-binary-manywheel-nightly.yml

@@ -38,7 +38,7 @@ jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
@@ -55,7 +55,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.9"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -63,7 +64,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_9-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_9-cpu-aarch64-test:  # Testing
@@ -79,7 +79,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.9"
       build_name: manywheel-py3_9-cpu-aarch64
@@ -102,7 +103,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.9"
       build_name: manywheel-py3_9-cpu-aarch64
@@ -122,7 +124,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.9"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -130,7 +133,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_9-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -148,7 +151,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.9"
       build_name: manywheel-py3_9-cuda-aarch64-12_8
@@ -167,7 +171,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.10"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -175,7 +180,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_10-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_10-cpu-aarch64-test:  # Testing
@@ -191,7 +195,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.10"
       build_name: manywheel-py3_10-cpu-aarch64
@@ -214,7 +219,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.10"
       build_name: manywheel-py3_10-cpu-aarch64
@@ -234,7 +240,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.10"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -242,7 +249,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_10-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -260,7 +267,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.10"
       build_name: manywheel-py3_10-cuda-aarch64-12_8
@@ -279,7 +287,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.11"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -287,7 +296,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_11-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_11-cpu-aarch64-test:  # Testing
@@ -303,7 +311,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.11"
       build_name: manywheel-py3_11-cpu-aarch64
@@ -326,7 +335,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.11"
       build_name: manywheel-py3_11-cpu-aarch64
@@ -346,7 +356,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.11"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -354,7 +365,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_11-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -372,7 +383,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.11"
       build_name: manywheel-py3_11-cuda-aarch64-12_8
@@ -391,7 +403,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.12"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -399,7 +412,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_12-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_12-cpu-aarch64-test:  # Testing
@@ -415,7 +427,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.12"
       build_name: manywheel-py3_12-cpu-aarch64
@@ -438,7 +451,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.12"
       build_name: manywheel-py3_12-cpu-aarch64
@@ -458,7 +472,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.12"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -466,7 +481,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_12-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -484,7 +499,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.12"
       build_name: manywheel-py3_12-cuda-aarch64-12_8
@@ -503,7 +519,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -511,7 +528,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13-cpu-aarch64-test:  # Testing
@@ -527,7 +543,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13"
       build_name: manywheel-py3_13-cpu-aarch64
@@ -550,7 +567,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13"
       build_name: manywheel-py3_13-cpu-aarch64
@@ -570,7 +588,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -578,7 +597,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -596,7 +615,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13"
       build_name: manywheel-py3_13-cuda-aarch64-12_8
@@ -615,7 +635,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13t"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -623,7 +644,6 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13t-cpu-aarch64
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.6.80; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.5.1.17; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.6.4.1; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.0.4; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.7.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.1.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.4.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.6.77; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.6.85; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.11.1.6; platform_system == 'Linux' and platform_machine == 'x86_64'
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
   manywheel-py3_13t-cpu-aarch64-test:  # Testing
@@ -639,7 +659,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13t"
       build_name: manywheel-py3_13t-cpu-aarch64
@@ -662,7 +683,8 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu-aarch64
-      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-main
+      DOCKER_IMAGE: pytorch/manylinux2_28_aarch64-builder:cpu-aarch64-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13t"
       build_name: manywheel-py3_13t-cpu-aarch64
@@ -682,7 +704,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13t"
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
@@ -690,7 +713,7 @@ jobs:
       ALPINE_IMAGE: "arm64v8/alpine"
       build_name: manywheel-py3_13t-cuda-aarch64-12_8
       build_environment: linux-aarch64-binary-manywheel
-      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.8.0.87; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
+      PYTORCH_EXTRA_INSTALL_REQUIREMENTS: nvidia-cuda-nvrtc-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-runtime-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cuda-cupti-cu12==12.8.57; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cudnn-cu12==9.7.1.26; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cublas-cu12==12.8.3.14; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufft-cu12==11.3.3.41; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-curand-cu12==10.3.9.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusolver-cu12==11.7.2.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparse-cu12==12.5.7.53; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cusparselt-cu12==0.6.3; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nccl-cu12==2.26.2; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvtx-cu12==12.8.55; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-nvjitlink-cu12==12.8.61; platform_system == 'Linux' and platform_machine == 'x86_64' | nvidia-cufile-cu12==1.13.0.11; platform_system == 'Linux' and platform_machine == 'x86_64'
       timeout-minutes: 420
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
@@ -708,7 +731,8 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8-aarch64
       GPU_ARCH_TYPE: cuda-aarch64
-      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-main
+      DOCKER_IMAGE: pytorch/manylinuxaarch64-builder:cuda12.8-2.7
+      DESIRED_DEVTOOLSET: cxx11-abi
       use_split_build: False
       DESIRED_PYTHON: "3.13t"
       build_name: manywheel-py3_13t-cuda-aarch64-12_8

.github/workflows/generated-linux-binary-libtorch-cxx11-abi-main.yml

@@ -2,7 +2,7 @@
 
 # Template is at:    .github/templates/linux_binary_build_workflow.yml.j2
 # Generation script: .github/scripts/generate_ci_workflows.py
-name: linux-binary-libtorch
+name: linux-binary-libtorch-cxx11-abi
 
 
 on:
@@ -18,7 +18,7 @@ env:
   ALPINE_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/tool/alpine"
   AWS_DEFAULT_REGION: us-east-1
   BINARY_ENV_FILE: /tmp/env
-  BUILD_ENVIRONMENT: linux-binary-libtorch
+  BUILD_ENVIRONMENT: linux-binary-libtorch-cxx11-abi
   GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
   PR_NUMBER: ${{ github.event.pull_request.number }}
   PYTORCH_FINAL_PACKAGE_DIR: /artifacts
@@ -26,20 +26,20 @@ env:
   SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
   SKIP_ALL_TESTS: 0
 concurrency:
-  group: linux-binary-libtorch-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  group: linux-binary-libtorch-cxx11-abi-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
   cancel-in-progress: true
 
 jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
       curr_branch: ${{ github.head_ref || github.ref_name }}
       curr_ref_type: ${{ github.ref_type }}
-  libtorch-cpu-shared-with-deps-release-build:
+  libtorch-cpu-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -50,18 +50,18 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-cpu-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-cpu-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cpu-shared-with-deps-release-test:  # Testing
+  libtorch-cpu-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-cpu-shared-with-deps-release-build
+      - libtorch-cpu-shared-with-deps-cxx11-abi-build
       - get-label-type
     uses: ./.github/workflows/_binary-test-linux.yml
     with:
@@ -71,11 +71,11 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cpu-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cpu-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       runs_on: linux.4xlarge
     secrets:

.github/workflows/generated-linux-binary-libtorch-cxx11-abi-nightly.yml

@@ -2,7 +2,7 @@
 
 # Template is at:    .github/templates/linux_binary_build_workflow.yml.j2
 # Generation script: .github/scripts/generate_ci_workflows.py
-name: linux-binary-libtorch
+name: linux-binary-libtorch-cxx11-abi
 
 
 on:
@@ -23,7 +23,7 @@ env:
   ALPINE_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/tool/alpine"
   AWS_DEFAULT_REGION: us-east-1
   BINARY_ENV_FILE: /tmp/env
-  BUILD_ENVIRONMENT: linux-binary-libtorch
+  BUILD_ENVIRONMENT: linux-binary-libtorch-cxx11-abi
   GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
   PR_NUMBER: ${{ github.event.pull_request.number }}
   PYTORCH_FINAL_PACKAGE_DIR: /artifacts
@@ -31,20 +31,20 @@ env:
   SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
   SKIP_ALL_TESTS: 0
 concurrency:
-  group: linux-binary-libtorch-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
+  group: linux-binary-libtorch-cxx11-abi-${{ github.event.pull_request.number || github.ref_name }}-${{ github.ref_type == 'branch' && github.sha }}-${{ github.event_name == 'workflow_dispatch' }}
   cancel-in-progress: true
 
 jobs:
   get-label-type:
     if: github.repository_owner == 'pytorch'
     name: get-label-type
-    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@main
+    uses: pytorch/pytorch/.github/workflows/_runner-determinator.yml@release/2.7
     with:
       triggering_actor: ${{ github.triggering_actor }}
       issue_owner: ${{ github.event.pull_request.user.login || github.event.issue.user.login }}
       curr_branch: ${{ github.head_ref || github.ref_name }}
       curr_ref_type: ${{ github.ref_type }}
-  libtorch-cpu-shared-with-deps-release-build:
+  libtorch-cpu-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -55,18 +55,18 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-cpu-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-cpu-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cpu-shared-with-deps-release-test:  # Testing
+  libtorch-cpu-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-cpu-shared-with-deps-release-build
+      - libtorch-cpu-shared-with-deps-cxx11-abi-build
       - get-label-type
     uses: ./.github/workflows/_binary-test-linux.yml
     with:
@@ -76,21 +76,21 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cpu-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cpu-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       runs_on: linux.4xlarge
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cpu-shared-with-deps-release-upload:  # Uploading
+  libtorch-cpu-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-cpu-shared-with-deps-release-test
+    needs: libtorch-cpu-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -98,15 +98,15 @@ jobs:
       #       favor of GPU_ARCH_VERSION
       DESIRED_CUDA: cpu
       GPU_ARCH_TYPE: cpu
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cpu-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cpu-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cpu-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml
 
-  libtorch-cuda11_8-shared-with-deps-release-build:
+  libtorch-cuda11_8-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -118,18 +118,18 @@ jobs:
       DESIRED_CUDA: cu118
       GPU_ARCH_VERSION: 11.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-cuda11_8-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-cuda11_8-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda11_8-shared-with-deps-release-test:  # Testing
+  libtorch-cuda11_8-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-cuda11_8-shared-with-deps-release-build
+      - libtorch-cuda11_8-shared-with-deps-cxx11-abi-build
       - get-label-type
     uses: ./.github/workflows/_binary-test-linux.yml
     with:
@@ -140,21 +140,21 @@ jobs:
       DESIRED_CUDA: cu118
       GPU_ARCH_VERSION: 11.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda11_8-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda11_8-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       runs_on: linux.4xlarge.nvidia.gpu
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda11_8-shared-with-deps-release-upload:  # Uploading
+  libtorch-cuda11_8-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-cuda11_8-shared-with-deps-release-test
+    needs: libtorch-cuda11_8-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -163,15 +163,15 @@ jobs:
       DESIRED_CUDA: cu118
       GPU_ARCH_VERSION: 11.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda11.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda11_8-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda11_8-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml
 
-  libtorch-cuda12_6-shared-with-deps-release-build:
+  libtorch-cuda12_6-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -183,18 +183,18 @@ jobs:
       DESIRED_CUDA: cu126
       GPU_ARCH_VERSION: 12.6
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-cuda12_6-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-cuda12_6-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda12_6-shared-with-deps-release-test:  # Testing
+  libtorch-cuda12_6-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-cuda12_6-shared-with-deps-release-build
+      - libtorch-cuda12_6-shared-with-deps-cxx11-abi-build
       - get-label-type
     uses: ./.github/workflows/_binary-test-linux.yml
     with:
@@ -205,21 +205,21 @@ jobs:
       DESIRED_CUDA: cu126
       GPU_ARCH_VERSION: 12.6
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda12_6-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda12_6-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       runs_on: linux.4xlarge.nvidia.gpu
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda12_6-shared-with-deps-release-upload:  # Uploading
+  libtorch-cuda12_6-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-cuda12_6-shared-with-deps-release-test
+    needs: libtorch-cuda12_6-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -228,15 +228,15 @@ jobs:
       DESIRED_CUDA: cu126
       GPU_ARCH_VERSION: 12.6
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.6-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda12_6-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda12_6-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml
 
-  libtorch-cuda12_8-shared-with-deps-release-build:
+  libtorch-cuda12_8-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -248,18 +248,18 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-cuda12_8-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-cuda12_8-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda12_8-shared-with-deps-release-test:  # Testing
+  libtorch-cuda12_8-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-cuda12_8-shared-with-deps-release-build
+      - libtorch-cuda12_8-shared-with-deps-cxx11-abi-build
       - get-label-type
     uses: ./.github/workflows/_binary-test-linux.yml
     with:
@@ -270,21 +270,21 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda12_8-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda12_8-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
       runs_on: linux.g4dn.4xlarge.nvidia.gpu  # 12.8 build needs sm_70+ runner
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-cuda12_8-shared-with-deps-release-upload:  # Uploading
+  libtorch-cuda12_8-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-cuda12_8-shared-with-deps-release-test
+    needs: libtorch-cuda12_8-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -293,15 +293,15 @@ jobs:
       DESIRED_CUDA: cu128
       GPU_ARCH_VERSION: 12.8
       GPU_ARCH_TYPE: cuda
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:cuda12.8-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-cuda12_8-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-cuda12_8-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml
 
-  libtorch-rocm6_2_4-shared-with-deps-release-build:
+  libtorch-rocm6_2_4-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -313,18 +313,18 @@ jobs:
       DESIRED_CUDA: rocm6.2.4
       GPU_ARCH_VERSION: 6.2.4
       GPU_ARCH_TYPE: rocm
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-rocm6_2_4-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-rocm6_2_4-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-rocm6_2_4-shared-with-deps-release-test:  # Testing
+  libtorch-rocm6_2_4-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-rocm6_2_4-shared-with-deps-release-build
+      - libtorch-rocm6_2_4-shared-with-deps-cxx11-abi-build
       - get-label-type
     runs-on: linux.rocm.gpu
     timeout-minutes: 240
@@ -337,21 +337,20 @@ jobs:
       GPU_ARCH_VERSION: 6.2.4
       GPU_ARCH_TYPE: rocm
       SKIP_ALL_TESTS: 1
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
     steps:
       - name: Setup ROCm
         uses: ./.github/actions/setup-rocm
       - uses: actions/download-artifact@v4.1.7
         name: Download Build Artifacts
         with:
-          name: libtorch-rocm6_2_4-shared-with-deps-release
+          name: libtorch-rocm6_2_4-shared-with-deps-cxx11-abi
           path: "${{ runner.temp }}/artifacts/"
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -364,19 +363,19 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
-          docker-image: pytorch/libtorch-cxx11-builder:rocm6.2.4-main
+          docker-image: pytorch/libtorch-cxx11-builder:rocm6.2.4-2.7
       - name: Test Pytorch binary
         uses: ./pytorch/.github/actions/test-pytorch-binary
       - name: Teardown ROCm
         uses: ./.github/actions/teardown-rocm
-  libtorch-rocm6_2_4-shared-with-deps-release-upload:  # Uploading
+  libtorch-rocm6_2_4-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-rocm6_2_4-shared-with-deps-release-test
+    needs: libtorch-rocm6_2_4-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -385,15 +384,15 @@ jobs:
       DESIRED_CUDA: rocm6.2.4
       GPU_ARCH_VERSION: 6.2.4
       GPU_ARCH_TYPE: rocm
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.2.4-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-rocm6_2_4-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-rocm6_2_4-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml
 
-  libtorch-rocm6_3-shared-with-deps-release-build:
+  libtorch-rocm6_3-shared-with-deps-cxx11-abi-build:
     if: ${{ github.repository_owner == 'pytorch' }}
     uses: ./.github/workflows/_binary-build-linux.yml
     needs: get-label-type
@@ -405,18 +404,18 @@ jobs:
       DESIRED_CUDA: rocm6.3
       GPU_ARCH_VERSION: 6.3
       GPU_ARCH_TYPE: rocm
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
       runner_prefix: "${{ needs.get-label-type.outputs.label-type }}"
-      build_name: libtorch-rocm6_3-shared-with-deps-release
-      build_environment: linux-binary-libtorch
+      build_name: libtorch-rocm6_3-shared-with-deps-cxx11-abi
+      build_environment: linux-binary-libtorch-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
-  libtorch-rocm6_3-shared-with-deps-release-test:  # Testing
+  libtorch-rocm6_3-shared-with-deps-cxx11-abi-test:  # Testing
     if: ${{ github.repository_owner == 'pytorch' }}
     needs:
-      - libtorch-rocm6_3-shared-with-deps-release-build
+      - libtorch-rocm6_3-shared-with-deps-cxx11-abi-build
       - get-label-type
     runs-on: linux.rocm.gpu
     timeout-minutes: 240
@@ -429,21 +428,20 @@ jobs:
       GPU_ARCH_VERSION: 6.3
       GPU_ARCH_TYPE: rocm
       SKIP_ALL_TESTS: 1
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-2.7
       LIBTORCH_VARIANT: shared-with-deps
+      DESIRED_DEVTOOLSET: cxx11-abi
     steps:
       - name: Setup ROCm
         uses: ./.github/actions/setup-rocm
       - uses: actions/download-artifact@v4.1.7
         name: Download Build Artifacts
         with:
-          name: libtorch-rocm6_3-shared-with-deps-release
+          name: libtorch-rocm6_3-shared-with-deps-cxx11-abi
           path: "${{ runner.temp }}/artifacts/"
       - name: Checkout PyTorch
         uses: actions/checkout@v4
         with:
-          ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
           submodules: recursive
           path: pytorch
           show-progress: false
@@ -456,19 +454,19 @@ jobs:
         run: |
           echo "GPU_FLAG=--device=/dev/mem --device=/dev/kfd --device=/dev/dri --group-add video --group-add daemon" >> "${GITHUB_ENV}"
       - name: Pull Docker image
-        uses: pytorch/test-infra/.github/actions/pull-docker-image@main
+        uses: pytorch/test-infra/.github/actions/pull-docker-image@release/2.7
         with:
-          docker-image: pytorch/libtorch-cxx11-builder:rocm6.3-main
+          docker-image: pytorch/libtorch-cxx11-builder:rocm6.3-2.7
       - name: Test Pytorch binary
         uses: ./pytorch/.github/actions/test-pytorch-binary
       - name: Teardown ROCm
         uses: ./.github/actions/teardown-rocm
-  libtorch-rocm6_3-shared-with-deps-release-upload:  # Uploading
+  libtorch-rocm6_3-shared-with-deps-cxx11-abi-upload:  # Uploading
     if: ${{ github.repository_owner == 'pytorch' }}
     permissions:
       id-token: write
       contents: read
-    needs: libtorch-rocm6_3-shared-with-deps-release-test
+    needs: libtorch-rocm6_3-shared-with-deps-cxx11-abi-test
     with:
       PYTORCH_ROOT: /pytorch
       PACKAGE_TYPE: libtorch
@@ -477,10 +475,10 @@ jobs:
       DESIRED_CUDA: rocm6.3
       GPU_ARCH_VERSION: 6.3
       GPU_ARCH_TYPE: rocm
-      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-main
-      LIBTORCH_CONFIG: release
+      DOCKER_IMAGE: pytorch/libtorch-cxx11-builder:rocm6.3-2.7
       LIBTORCH_VARIANT: shared-with-deps
-      build_name: libtorch-rocm6_3-shared-with-deps-release
+      DESIRED_DEVTOOLSET: cxx11-abi
+      build_name: libtorch-rocm6_3-shared-with-deps-cxx11-abi
     secrets:
       github-token: ${{ secrets.GITHUB_TOKEN }}
     uses: ./.github/workflows/_binary-upload.yml