From a736c41187363d39cf72d1b4a3bda5b62554f3a1 Mon Sep 17 00:00:00 2001
From: "Sun, Diwei" <diwei.sun@intel.com>
Date: Tue, 19 Aug 2025 08:28:02 +0000
Subject: [PATCH 01/25] enable xpu ci test

---
 .github/workflows/pr-test-xpu.yml | 156 ++++++++++++++++++++++++++++++
 1 file changed, 156 insertions(+)
 create mode 100644 .github/workflows/pr-test-xpu.yml

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
new file mode 100644
index 0000000000..79621a06d1
--- /dev/null
+++ b/.github/workflows/pr-test-xpu.yml
@@ -0,0 +1,156 @@
+# TODO: this looks sort of similar to _linux-test, but there are like a dozen
+# places where you would have to insert an if statement. Probably it's better to
+# just use a different workflow altogether
+
+name: xpu-test
+
+on:
+  push:
+    branches:
+      - main
+      - 'gh/**'
+  pull_request:
+    branches:
+      - main
+      - 'gh/**'
+
+concurrency:
+  group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  test:
+    # Don't run on forked repos or empty test matrix
+    # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
+    timeout-minutes: 60
+    runs-on: ao-pvc
+    env:
+      DOCKER_IMAGE: ghcr.io/pytorch/ci-image:pytorch-linux-jammy-xpu-2025.1-py3-b388c12018df5d6ce2f94b7fb337fa3729978ab3
+      TEST_COMMAND: .github/scripts/ci_test_xpu.sh
+      PYTORCH_RETRY_TEST_CASES: 1
+      PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
+      XLA_CLANG_CACHE_S3_BUCKET_NAME: ossci-compiler-clang-cache-circleci-xla
+    steps:
+      # [see note: pytorch repo ref]
+      - name: Checkout Torchao
+        uses: actions/checkout@v4
+
+      - name: Clean all stopped docker containers
+        if: always()
+        shell: bash
+        run: |
+          # Prune all stopped containers.
+          # If other runner is pruning on this node, will skip.
+          nprune=$(ps -ef | grep -c "docker container prune")
+          if [[ $nprune -eq 1 ]]; then
+            docker container prune -f
+          fi
+
+      - name: Runner health check GPU count
+        if: always()
+        shell: bash
+        run: |
+          ngpu=$(timeout 30 clinfo -l | grep -c -E 'Device' || true)
+          msg="Please file an issue on pytorch/ao reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
+          if [[ $ngpu -eq 0 ]]; then
+            echo "Error: Failed to detect any GPUs on the runner"
+            echo "$msg"
+            exit 1
+          fi
+
+      - name: Use following to pull public copy of the image
+        id: print-ghcr-mirror
+        shell: bash
+        run: |
+          echo "docker pull ${DOCKER_IMAGE}"
+          docker pull ${DOCKER_IMAGE}
+
+      - name: Test
+        id: test
+        env:
+          BUILD_ENVIRONMENT: ${{ inputs.build-environment }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          GITHUB_REPOSITORY: ${{ github.repository }}
+          GITHUB_WORKFLOW: ${{ github.workflow }}
+          GITHUB_JOB: ${{ github.job }}
+          GITHUB_RUN_ID: ${{ github.run_id }}
+          GITHUB_RUN_NUMBER: ${{ github.run_number }}
+          GITHUB_RUN_ATTEMPT: ${{ github.run_attempt }}
+          SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
+        timeout-minutes: 60
+        run: |
+          set -x
+
+          # detached container should get cleaned up by teardown_ec2_linux
+          # Used for GPU_FLAG since that doesn't play nice
+          # shellcheck disable=SC2086,SC2090
+          container_name=$(docker run \
+            ${GPU_FLAG:-} \
+            -e BUILD_ENVIRONMENT \
+            -e PR_NUMBER \
+            -e GITHUB_ACTIONS \
+            -e GITHUB_REPOSITORY \
+            -e GITHUB_WORKFLOW \
+            -e GITHUB_JOB \
+            -e GITHUB_RUN_ID \
+            -e GITHUB_RUN_NUMBER \
+            -e GITHUB_RUN_ATTEMPT \
+            -e JOB_ID \
+            -e BRANCH \
+            -e SHA1 \
+            --user $(id -u):$(id -g) \
+            --ulimit stack=10485760:83886080 \
+            --ulimit core=0 \
+            --security-opt seccomp=unconfined \
+            --cap-add=SYS_PTRACE \
+            --shm-size="8g" \
+            --tty \
+            --detach \
+            --name="${container_name}" \
+            --user jenkins \
+            --privileged \
+            -v "${GITHUB_WORKSPACE}:/var/lib/jenkins/workspace" \
+            -w /var/lib/jenkins/workspace \
+            "${DOCKER_IMAGE}"
+          )
+          # save container name for later step
+          echo "CONTAINER_NAME=${container_name}" >> "$GITHUB_ENV"
+          # jenkins user does not have write permission to mounted workspace; work-around by copying within container to jenkins home
+          docker exec -t "${container_name}" sh -c "bash ${TEST_COMMAND}"
+
+      - name: Change permissions
+        if: ${{ always() && steps.test.conclusion }}
+        run: |
+          docker exec -t "${{ env.CONTAINER_NAME }}" sh -c "sudo chown -R 1001:1001 test"
+
+      - name: Collect backtraces from coredumps (if any)
+        if: always()
+        run: |
+          # shellcheck disable=SC2156
+          find . -iname "core.[1-9]*" -exec docker exec "${CONTAINER_NAME}" sh -c "gdb python {} -ex 'bt' -ex 'q'" \;
+
+      - name: Stop container before exit
+        if: always()
+        run: |
+          # Workaround for multiple runners on same IDC node
+          docker stop "${{ env.CONTAINER_NAME }}"
+
+      - name: Store Core dumps on GitHub
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
+        if: failure()
+        with:
+          name: coredumps-${{ matrix.config }}-${{ matrix.shard }}-${{ matrix.num_shards }}-${{ matrix.runner }}
+          retention-days: 14
+          if-no-files-found: ignore
+          path: ./**/core.[1-9]*
+
+      - name: Teardown XPU
+        if: always()
+        shell: bash
+        run: |
+          # Prune all stopped containers.
+          # If other runner is pruning on this node, will skip.
+          nprune=$(ps -ef | grep -c "docker container prune")
+          if [[ $nprune -eq 1 ]]; then
+            docker container prune -f
+          fi

From 7c96ad46d16039d37a685fa9ed28010d1a3ab0aa Mon Sep 17 00:00:00 2001
From: "Sun, Diwei" <diwei.sun@intel.com>
Date: Wed, 20 Aug 2025 01:52:54 +0000
Subject: [PATCH 02/25] Revert "enable xpu ci test"

This reverts commit a736c41187363d39cf72d1b4a3bda5b62554f3a1.
---
 .github/workflows/pr-test-xpu.yml | 156 ------------------------------
 1 file changed, 156 deletions(-)
 delete mode 100644 .github/workflows/pr-test-xpu.yml

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
deleted file mode 100644
index 79621a06d1..0000000000
--- a/.github/workflows/pr-test-xpu.yml
+++ /dev/null
@@ -1,156 +0,0 @@
-# TODO: this looks sort of similar to _linux-test, but there are like a dozen
-# places where you would have to insert an if statement. Probably it's better to
-# just use a different workflow altogether
-
-name: xpu-test
-
-on:
-  push:
-    branches:
-      - main
-      - 'gh/**'
-  pull_request:
-    branches:
-      - main
-      - 'gh/**'
-
-concurrency:
-  group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  test:
-    # Don't run on forked repos or empty test matrix
-    # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
-    timeout-minutes: 60
-    runs-on: ao-pvc
-    env:
-      DOCKER_IMAGE: ghcr.io/pytorch/ci-image:pytorch-linux-jammy-xpu-2025.1-py3-b388c12018df5d6ce2f94b7fb337fa3729978ab3
-      TEST_COMMAND: .github/scripts/ci_test_xpu.sh
-      PYTORCH_RETRY_TEST_CASES: 1
-      PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
-      XLA_CLANG_CACHE_S3_BUCKET_NAME: ossci-compiler-clang-cache-circleci-xla
-    steps:
-      # [see note: pytorch repo ref]
-      - name: Checkout Torchao
-        uses: actions/checkout@v4
-
-      - name: Clean all stopped docker containers
-        if: always()
-        shell: bash
-        run: |
-          # Prune all stopped containers.
-          # If other runner is pruning on this node, will skip.
-          nprune=$(ps -ef | grep -c "docker container prune")
-          if [[ $nprune -eq 1 ]]; then
-            docker container prune -f
-          fi
-
-      - name: Runner health check GPU count
-        if: always()
-        shell: bash
-        run: |
-          ngpu=$(timeout 30 clinfo -l | grep -c -E 'Device' || true)
-          msg="Please file an issue on pytorch/ao reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
-          if [[ $ngpu -eq 0 ]]; then
-            echo "Error: Failed to detect any GPUs on the runner"
-            echo "$msg"
-            exit 1
-          fi
-
-      - name: Use following to pull public copy of the image
-        id: print-ghcr-mirror
-        shell: bash
-        run: |
-          echo "docker pull ${DOCKER_IMAGE}"
-          docker pull ${DOCKER_IMAGE}
-
-      - name: Test
-        id: test
-        env:
-          BUILD_ENVIRONMENT: ${{ inputs.build-environment }}
-          PR_NUMBER: ${{ github.event.pull_request.number }}
-          GITHUB_REPOSITORY: ${{ github.repository }}
-          GITHUB_WORKFLOW: ${{ github.workflow }}
-          GITHUB_JOB: ${{ github.job }}
-          GITHUB_RUN_ID: ${{ github.run_id }}
-          GITHUB_RUN_NUMBER: ${{ github.run_number }}
-          GITHUB_RUN_ATTEMPT: ${{ github.run_attempt }}
-          SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
-        timeout-minutes: 60
-        run: |
-          set -x
-
-          # detached container should get cleaned up by teardown_ec2_linux
-          # Used for GPU_FLAG since that doesn't play nice
-          # shellcheck disable=SC2086,SC2090
-          container_name=$(docker run \
-            ${GPU_FLAG:-} \
-            -e BUILD_ENVIRONMENT \
-            -e PR_NUMBER \
-            -e GITHUB_ACTIONS \
-            -e GITHUB_REPOSITORY \
-            -e GITHUB_WORKFLOW \
-            -e GITHUB_JOB \
-            -e GITHUB_RUN_ID \
-            -e GITHUB_RUN_NUMBER \
-            -e GITHUB_RUN_ATTEMPT \
-            -e JOB_ID \
-            -e BRANCH \
-            -e SHA1 \
-            --user $(id -u):$(id -g) \
-            --ulimit stack=10485760:83886080 \
-            --ulimit core=0 \
-            --security-opt seccomp=unconfined \
-            --cap-add=SYS_PTRACE \
-            --shm-size="8g" \
-            --tty \
-            --detach \
-            --name="${container_name}" \
-            --user jenkins \
-            --privileged \
-            -v "${GITHUB_WORKSPACE}:/var/lib/jenkins/workspace" \
-            -w /var/lib/jenkins/workspace \
-            "${DOCKER_IMAGE}"
-          )
-          # save container name for later step
-          echo "CONTAINER_NAME=${container_name}" >> "$GITHUB_ENV"
-          # jenkins user does not have write permission to mounted workspace; work-around by copying within container to jenkins home
-          docker exec -t "${container_name}" sh -c "bash ${TEST_COMMAND}"
-
-      - name: Change permissions
-        if: ${{ always() && steps.test.conclusion }}
-        run: |
-          docker exec -t "${{ env.CONTAINER_NAME }}" sh -c "sudo chown -R 1001:1001 test"
-
-      - name: Collect backtraces from coredumps (if any)
-        if: always()
-        run: |
-          # shellcheck disable=SC2156
-          find . -iname "core.[1-9]*" -exec docker exec "${CONTAINER_NAME}" sh -c "gdb python {} -ex 'bt' -ex 'q'" \;
-
-      - name: Stop container before exit
-        if: always()
-        run: |
-          # Workaround for multiple runners on same IDC node
-          docker stop "${{ env.CONTAINER_NAME }}"
-
-      - name: Store Core dumps on GitHub
-        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
-        if: failure()
-        with:
-          name: coredumps-${{ matrix.config }}-${{ matrix.shard }}-${{ matrix.num_shards }}-${{ matrix.runner }}
-          retention-days: 14
-          if-no-files-found: ignore
-          path: ./**/core.[1-9]*
-
-      - name: Teardown XPU
-        if: always()
-        shell: bash
-        run: |
-          # Prune all stopped containers.
-          # If other runner is pruning on this node, will skip.
-          nprune=$(ps -ef | grep -c "docker container prune")
-          if [[ $nprune -eq 1 ]]; then
-            docker container prune -f
-          fi

From d1122fcd9c6ab74aaa6cf2c4deb183c08e5360a1 Mon Sep 17 00:00:00 2001
From: "Sun, Diwei" <diwei.sun@intel.com>
Date: Wed, 20 Aug 2025 01:55:27 +0000
Subject: [PATCH 03/25] enable ci test for xpu

---
 .github/workflows/pr-test-xpu.yml | 156 ++++++++++++++++++++++++++++++
 1 file changed, 156 insertions(+)
 create mode 100644 .github/workflows/pr-test-xpu.yml

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
new file mode 100644
index 0000000000..79621a06d1
--- /dev/null
+++ b/.github/workflows/pr-test-xpu.yml
@@ -0,0 +1,156 @@
+# TODO: this looks sort of similar to _linux-test, but there are like a dozen
+# places where you would have to insert an if statement. Probably it's better to
+# just use a different workflow altogether
+
+name: xpu-test
+
+on:
+  push:
+    branches:
+      - main
+      - 'gh/**'
+  pull_request:
+    branches:
+      - main
+      - 'gh/**'
+
+concurrency:
+  group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  test:
+    # Don't run on forked repos or empty test matrix
+    # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
+    timeout-minutes: 60
+    runs-on: ao-pvc
+    env:
+      DOCKER_IMAGE: ghcr.io/pytorch/ci-image:pytorch-linux-jammy-xpu-2025.1-py3-b388c12018df5d6ce2f94b7fb337fa3729978ab3
+      TEST_COMMAND: .github/scripts/ci_test_xpu.sh
+      PYTORCH_RETRY_TEST_CASES: 1
+      PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
+      XLA_CLANG_CACHE_S3_BUCKET_NAME: ossci-compiler-clang-cache-circleci-xla
+    steps:
+      # [see note: pytorch repo ref]
+      - name: Checkout Torchao
+        uses: actions/checkout@v4
+
+      - name: Clean all stopped docker containers
+        if: always()
+        shell: bash
+        run: |
+          # Prune all stopped containers.
+          # If other runner is pruning on this node, will skip.
+          nprune=$(ps -ef | grep -c "docker container prune")
+          if [[ $nprune -eq 1 ]]; then
+            docker container prune -f
+          fi
+
+      - name: Runner health check GPU count
+        if: always()
+        shell: bash
+        run: |
+          ngpu=$(timeout 30 clinfo -l | grep -c -E 'Device' || true)
+          msg="Please file an issue on pytorch/ao reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
+          if [[ $ngpu -eq 0 ]]; then
+            echo "Error: Failed to detect any GPUs on the runner"
+            echo "$msg"
+            exit 1
+          fi
+
+      - name: Use following to pull public copy of the image
+        id: print-ghcr-mirror
+        shell: bash
+        run: |
+          echo "docker pull ${DOCKER_IMAGE}"
+          docker pull ${DOCKER_IMAGE}
+
+      - name: Test
+        id: test
+        env:
+          BUILD_ENVIRONMENT: ${{ inputs.build-environment }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          GITHUB_REPOSITORY: ${{ github.repository }}
+          GITHUB_WORKFLOW: ${{ github.workflow }}
+          GITHUB_JOB: ${{ github.job }}
+          GITHUB_RUN_ID: ${{ github.run_id }}
+          GITHUB_RUN_NUMBER: ${{ github.run_number }}
+          GITHUB_RUN_ATTEMPT: ${{ github.run_attempt }}
+          SHA1: ${{ github.event.pull_request.head.sha || github.sha }}
+        timeout-minutes: 60
+        run: |
+          set -x
+
+          # detached container should get cleaned up by teardown_ec2_linux
+          # Used for GPU_FLAG since that doesn't play nice
+          # shellcheck disable=SC2086,SC2090
+          container_name=$(docker run \
+            ${GPU_FLAG:-} \
+            -e BUILD_ENVIRONMENT \
+            -e PR_NUMBER \
+            -e GITHUB_ACTIONS \
+            -e GITHUB_REPOSITORY \
+            -e GITHUB_WORKFLOW \
+            -e GITHUB_JOB \
+            -e GITHUB_RUN_ID \
+            -e GITHUB_RUN_NUMBER \
+            -e GITHUB_RUN_ATTEMPT \
+            -e JOB_ID \
+            -e BRANCH \
+            -e SHA1 \
+            --user $(id -u):$(id -g) \
+            --ulimit stack=10485760:83886080 \
+            --ulimit core=0 \
+            --security-opt seccomp=unconfined \
+            --cap-add=SYS_PTRACE \
+            --shm-size="8g" \
+            --tty \
+            --detach \
+            --name="${container_name}" \
+            --user jenkins \
+            --privileged \
+            -v "${GITHUB_WORKSPACE}:/var/lib/jenkins/workspace" \
+            -w /var/lib/jenkins/workspace \
+            "${DOCKER_IMAGE}"
+          )
+          # save container name for later step
+          echo "CONTAINER_NAME=${container_name}" >> "$GITHUB_ENV"
+          # jenkins user does not have write permission to mounted workspace; work-around by copying within container to jenkins home
+          docker exec -t "${container_name}" sh -c "bash ${TEST_COMMAND}"
+
+      - name: Change permissions
+        if: ${{ always() && steps.test.conclusion }}
+        run: |
+          docker exec -t "${{ env.CONTAINER_NAME }}" sh -c "sudo chown -R 1001:1001 test"
+
+      - name: Collect backtraces from coredumps (if any)
+        if: always()
+        run: |
+          # shellcheck disable=SC2156
+          find . -iname "core.[1-9]*" -exec docker exec "${CONTAINER_NAME}" sh -c "gdb python {} -ex 'bt' -ex 'q'" \;
+
+      - name: Stop container before exit
+        if: always()
+        run: |
+          # Workaround for multiple runners on same IDC node
+          docker stop "${{ env.CONTAINER_NAME }}"
+
+      - name: Store Core dumps on GitHub
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
+        if: failure()
+        with:
+          name: coredumps-${{ matrix.config }}-${{ matrix.shard }}-${{ matrix.num_shards }}-${{ matrix.runner }}
+          retention-days: 14
+          if-no-files-found: ignore
+          path: ./**/core.[1-9]*
+
+      - name: Teardown XPU
+        if: always()
+        shell: bash
+        run: |
+          # Prune all stopped containers.
+          # If other runner is pruning on this node, will skip.
+          nprune=$(ps -ef | grep -c "docker container prune")
+          if [[ $nprune -eq 1 ]]; then
+            docker container prune -f
+          fi

From 4593d9534bd1ed9e1d3c590b4b8caa63cbf397b5 Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Wed, 20 Aug 2025 14:50:59 +0800
Subject: [PATCH 04/25] Create ci_test_xpu.sh

---
 .github/scripts/ci_test_xpu.sh | 9 +++++++++
 1 file changed, 9 insertions(+)
 create mode 100644 .github/scripts/ci_test_xpu.sh

diff --git a/.github/scripts/ci_test_xpu.sh b/.github/scripts/ci_test_xpu.sh
new file mode 100644
index 0000000000..ccff1b848f
--- /dev/null
+++ b/.github/scripts/ci_test_xpu.sh
@@ -0,0 +1,9 @@
+#!/bin/bash
+
+python3 -m pip install torch torchvision torchaudio pytorch-triton-xpu --index-url https://download.pytorch.org/whl/nightly/xpu --force-reinstall --no-cache-dir 
+python3 setup.py install
+
+pip install pytest expecttest parameterized accelerate hf_transfer 'modelscope!=1.15.0'
+
+cd test/quantization
+pytest -v -s *.py

From 7d90b8c923747e4f3897c8f42517ea5c49ebdafd Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Fri, 22 Aug 2025 14:30:44 +0800
Subject: [PATCH 05/25] Update .github/workflows/pr-test-xpu.yml

Co-authored-by: Wang, Chuanqi <chuanqi.wang@intel.com>
---
 .github/workflows/pr-test-xpu.yml | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 79621a06d1..72c33bdef5 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -11,8 +11,7 @@ on:
       - 'gh/**'
   pull_request:
     branches:
-      - main
-      - 'gh/**'
+      - ciflow/xpu/*
 
 concurrency:
   group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}

From e6bc407a053ad939cb14bf92baa32f0303b60f70 Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Fri, 22 Aug 2025 14:30:54 +0800
Subject: [PATCH 06/25] Update .github/workflows/pr-test-xpu.yml

Co-authored-by: Wang, Chuanqi <chuanqi.wang@intel.com>
---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 72c33bdef5..f18975e2eb 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -22,7 +22,7 @@ jobs:
     # Don't run on forked repos or empty test matrix
     # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
     timeout-minutes: 60
-    runs-on: ao-pvc
+    runs-on: linux.idc.xpu
     env:
       DOCKER_IMAGE: ghcr.io/pytorch/ci-image:pytorch-linux-jammy-xpu-2025.1-py3-b388c12018df5d6ce2f94b7fb337fa3729978ab3
       TEST_COMMAND: .github/scripts/ci_test_xpu.sh

From c34601f553c1c810ea34d852e187feadc0ee0efc Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:00:53 +0800
Subject: [PATCH 07/25] fix for trigger scenarios

---
 .github/workflows/pr-test-xpu.yml | 1 -
 1 file changed, 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index f18975e2eb..9670d1891d 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -8,7 +8,6 @@ on:
   push:
     branches:
       - main
-      - 'gh/**'
   pull_request:
     branches:
       - ciflow/xpu/*

From 3085c2b27a29ce461faaf33260f77c34ce399882 Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:53:50 +0800
Subject: [PATCH 08/25] port from pytorch repo

---
 .github/workflows/action.yml | 67 ++++++++++++++++++++++++++++++++++++
 1 file changed, 67 insertions(+)
 create mode 100644 .github/workflows/action.yml

diff --git a/.github/workflows/action.yml b/.github/workflows/action.yml
new file mode 100644
index 0000000000..740492475d
--- /dev/null
+++ b/.github/workflows/action.yml
@@ -0,0 +1,67 @@
+name: Setup XPU host
+
+description: Set up XPU host for CI
+
+runs:
+  using: composite
+  steps:
+    - name: Clean all stopped docker containers
+      if: always()
+      shell: bash
+      run: |
+        # Prune all stopped containers.
+        # If other runner is pruning on this node, will skip.
+        nprune=$(ps -ef | grep -c "docker container prune")
+        if [[ $nprune -eq 1 ]]; then
+          docker container prune -f
+        fi
+
+    - name: Runner health check system info
+      if: always()
+      shell: bash
+      run: |
+        cat /etc/os-release || true
+        cat /etc/apt/sources.list.d/oneAPI.list || true
+        cat /etc/apt/sources.list.d/intel-gpu-jammy.list || true
+        whoami
+
+    - name: Runner health check xpu-smi
+      if: always()
+      shell: bash
+      run: |
+        timeout 30 xpu-smi discovery || true
+
+    - name: Runner health check GPU count
+      if: always()
+      shell: bash
+      run: |
+        ngpu=$(timeout 30 xpu-smi discovery | grep -c -E 'Device Name' || true)
+        msg="Please file an issue on pytorch/pytorch reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
+        if [[ $ngpu -eq 0 ]]; then
+          echo "Error: Failed to detect any GPUs on the runner"
+          echo "$msg"
+          exit 1
+        fi
+
+    - name: Runner diskspace health check
+      uses: ./.github/actions/diskspace-cleanup
+      if: always()
+
+    - name: Runner health check disconnect on failure
+      if: ${{ failure() }}
+      shell: bash
+      run: |
+        killall runsvc.sh
+
+    - name: Preserve github env variables for use in docker
+      shell: bash
+      run: |
+        env | grep '^GITHUB' >> "/tmp/github_env_${GITHUB_RUN_ID}"
+        env | grep '^CI' >> "/tmp/github_env_${GITHUB_RUN_ID}"
+
+    - name: XPU set GPU_FLAG
+      shell: bash
+      run: |
+        # Add render group for container creation.
+        render_gid=`cat /etc/group | grep render | cut -d: -f3`
+        echo "GPU_FLAG=--device=/dev/mem --device=/dev/dri --group-add video --group-add $render_gid" >> "${GITHUB_ENV}"

From d9ab09e8a9ad936745a402628d305ef1bad9b854 Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:55:00 +0800
Subject: [PATCH 09/25] Rename action.yml to xpu-action.yml

---
 .github/workflows/{action.yml => xpu-action.yml} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename .github/workflows/{action.yml => xpu-action.yml} (100%)

diff --git a/.github/workflows/action.yml b/.github/workflows/xpu-action.yml
similarity index 100%
rename from .github/workflows/action.yml
rename to .github/workflows/xpu-action.yml

From f87892abcebacc44e71a837eb5ee4a21edf6e6a9 Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Tue, 2 Sep 2025 10:59:07 +0800
Subject: [PATCH 10/25] update to align with pytorch

---
 .github/workflows/pr-test-xpu.yml | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 9670d1891d..75991289e1 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -23,7 +23,7 @@ jobs:
     timeout-minutes: 60
     runs-on: linux.idc.xpu
     env:
-      DOCKER_IMAGE: ghcr.io/pytorch/ci-image:pytorch-linux-jammy-xpu-2025.1-py3-b388c12018df5d6ce2f94b7fb337fa3729978ab3
+      DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-1-py3
       TEST_COMMAND: .github/scripts/ci_test_xpu.sh
       PYTORCH_RETRY_TEST_CASES: 1
       PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
@@ -33,6 +33,20 @@ jobs:
       - name: Checkout Torchao
         uses: actions/checkout@v4
 
+      - name: Setup XPU
+        uses: ./.github/workflows/xpu-action.yml
+
+      - name: configure aws credentials
+        id: aws_creds
+        uses: aws-actions/configure-aws-credentials@ececac1a45f3b08a01d2dd070d28d111c5fe6722 # v4.1.0
+        with:
+          role-to-assume: arn:aws:iam::308535385114:role/gha_workflow_s3_and_ecr_read_only
+          aws-region: us-east-1
+
+      - name: Login to Amazon ECR
+        id: login-ecr
+        uses: aws-actions/amazon-ecr-login@062b18b96a7aff071d4dc91bc00c4c1a7945b076 # v2.0.1
+      
       - name: Clean all stopped docker containers
         if: always()
         shell: bash

From c6f07b5aa1c2d9dd40151b365b56a6e89678792c Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Wed, 3 Sep 2025 19:40:04 -0700
Subject: [PATCH 11/25] Revert "Rename action.yml to xpu-action.yml"

This reverts commit d9ab09e8a9ad936745a402628d305ef1bad9b854.
---
 .github/workflows/{xpu-action.yml => action.yml} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename .github/workflows/{xpu-action.yml => action.yml} (100%)

diff --git a/.github/workflows/xpu-action.yml b/.github/workflows/action.yml
similarity index 100%
rename from .github/workflows/xpu-action.yml
rename to .github/workflows/action.yml

From 544593a5a1a7f54d9c7792d6ae62c9dca6f891bd Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Wed, 3 Sep 2025 19:40:13 -0700
Subject: [PATCH 12/25] Revert "port from pytorch repo"

This reverts commit 3085c2b27a29ce461faaf33260f77c34ce399882.
---
 .github/workflows/action.yml | 67 ------------------------------------
 1 file changed, 67 deletions(-)
 delete mode 100644 .github/workflows/action.yml

diff --git a/.github/workflows/action.yml b/.github/workflows/action.yml
deleted file mode 100644
index 740492475d..0000000000
--- a/.github/workflows/action.yml
+++ /dev/null
@@ -1,67 +0,0 @@
-name: Setup XPU host
-
-description: Set up XPU host for CI
-
-runs:
-  using: composite
-  steps:
-    - name: Clean all stopped docker containers
-      if: always()
-      shell: bash
-      run: |
-        # Prune all stopped containers.
-        # If other runner is pruning on this node, will skip.
-        nprune=$(ps -ef | grep -c "docker container prune")
-        if [[ $nprune -eq 1 ]]; then
-          docker container prune -f
-        fi
-
-    - name: Runner health check system info
-      if: always()
-      shell: bash
-      run: |
-        cat /etc/os-release || true
-        cat /etc/apt/sources.list.d/oneAPI.list || true
-        cat /etc/apt/sources.list.d/intel-gpu-jammy.list || true
-        whoami
-
-    - name: Runner health check xpu-smi
-      if: always()
-      shell: bash
-      run: |
-        timeout 30 xpu-smi discovery || true
-
-    - name: Runner health check GPU count
-      if: always()
-      shell: bash
-      run: |
-        ngpu=$(timeout 30 xpu-smi discovery | grep -c -E 'Device Name' || true)
-        msg="Please file an issue on pytorch/pytorch reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
-        if [[ $ngpu -eq 0 ]]; then
-          echo "Error: Failed to detect any GPUs on the runner"
-          echo "$msg"
-          exit 1
-        fi
-
-    - name: Runner diskspace health check
-      uses: ./.github/actions/diskspace-cleanup
-      if: always()
-
-    - name: Runner health check disconnect on failure
-      if: ${{ failure() }}
-      shell: bash
-      run: |
-        killall runsvc.sh
-
-    - name: Preserve github env variables for use in docker
-      shell: bash
-      run: |
-        env | grep '^GITHUB' >> "/tmp/github_env_${GITHUB_RUN_ID}"
-        env | grep '^CI' >> "/tmp/github_env_${GITHUB_RUN_ID}"
-
-    - name: XPU set GPU_FLAG
-      shell: bash
-      run: |
-        # Add render group for container creation.
-        render_gid=`cat /etc/group | grep render | cut -d: -f3`
-        echo "GPU_FLAG=--device=/dev/mem --device=/dev/dri --group-add video --group-add $render_gid" >> "${GITHUB_ENV}"

From 2e1dc508c12ddcdbc0bec7649984d2306db356be Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Thu, 4 Sep 2025 14:43:22 +0800
Subject: [PATCH 13/25] Update .github/workflows/pr-test-xpu.yml

Co-authored-by: Wang, Chuanqi <chuanqi.wang@intel.com>
---
 .github/workflows/pr-test-xpu.yml | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 75991289e1..f4e27832ad 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -6,11 +6,11 @@ name: xpu-test
 
 on:
   push:
-    branches:
-      - main
+    tags:
+      - ciflow/xpu/*
   pull_request:
     branches:
-      - ciflow/xpu/*
+      - main
 
 concurrency:
   group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}

From 188a0f8bf0508efd3a24909718b078d4ca974c3c Mon Sep 17 00:00:00 2001
From: DiweiSun <105627594+DiweiSun@users.noreply.github.com>
Date: Fri, 5 Sep 2025 09:27:40 +0800
Subject: [PATCH 14/25] debug for runner

---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index f4e27832ad..25531166a6 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -21,7 +21,7 @@ jobs:
     # Don't run on forked repos or empty test matrix
     # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
     timeout-minutes: 60
-    runs-on: linux.idc.xpu
+    runs-on: ubuntu-latest
     env:
       DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-1-py3
       TEST_COMMAND: .github/scripts/ci_test_xpu.sh

From 421d02cfdac942e66c3ce1c72593cbcbd06ac1d2 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Mon, 8 Sep 2025 00:58:02 -0700
Subject: [PATCH 15/25] lint format fix

---
 .github/workflows/pr-test-xpu.yml | 41 ++++++++++++++++---------------
 1 file changed, 21 insertions(+), 20 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 25531166a6..6c36b32233 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -23,7 +23,7 @@ jobs:
     timeout-minutes: 60
     runs-on: ubuntu-latest
     env:
-      DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-1-py3
+      DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-py3
       TEST_COMMAND: .github/scripts/ci_test_xpu.sh
       PYTORCH_RETRY_TEST_CASES: 1
       PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
@@ -34,7 +34,7 @@ jobs:
         uses: actions/checkout@v4
 
       - name: Setup XPU
-        uses: ./.github/workflows/xpu-action.yml
+        uses: pytorch/pytorch/.github/actions/setup-xpu@main
 
       - name: configure aws credentials
         id: aws_creds
@@ -46,17 +46,26 @@ jobs:
       - name: Login to Amazon ECR
         id: login-ecr
         uses: aws-actions/amazon-ecr-login@062b18b96a7aff071d4dc91bc00c4c1a7945b076 # v2.0.1
-      
-      - name: Clean all stopped docker containers
-        if: always()
+
+      - name: Calculate docker image
+        id: calculate-docker-image
+        uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
+        with:
+          docker-image-name: ${{ DOCKER_IMAGE }}
+
+      - name: Use following to pull public copy of the image
+        id: print-ghcr-mirror
+        env:
+          ECR_DOCKER_IMAGE: ${{ steps.calculate-docker-image.outputs.docker-image }}
         shell: bash
         run: |
-          # Prune all stopped containers.
-          # If other runner is pruning on this node, will skip.
-          nprune=$(ps -ef | grep -c "docker container prune")
-          if [[ $nprune -eq 1 ]]; then
-            docker container prune -f
-          fi
+          tag=${ECR_DOCKER_IMAGE##*:}
+          echo "docker pull ghcr.io/pytorch/ci-image:${tag/:/-}"
+
+      - 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: Runner health check GPU count
         if: always()
@@ -157,12 +166,4 @@ jobs:
           path: ./**/core.[1-9]*
 
       - name: Teardown XPU
-        if: always()
-        shell: bash
-        run: |
-          # Prune all stopped containers.
-          # If other runner is pruning on this node, will skip.
-          nprune=$(ps -ef | grep -c "docker container prune")
-          if [[ $nprune -eq 1 ]]; then
-            docker container prune -f
-          fi
+        uses: pytorch/pytorch/.github/actions/teardown-xpu

From 6f6cd17dec5e3288993fc2c0b04217bf5eff6968 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Mon, 8 Sep 2025 01:13:38 -0700
Subject: [PATCH 16/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 16 ++++------------
 1 file changed, 4 insertions(+), 12 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 6c36b32233..d4ab4e43d5 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -24,7 +24,6 @@ jobs:
     runs-on: ubuntu-latest
     env:
       DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-py3
-      TEST_COMMAND: .github/scripts/ci_test_xpu.sh
       PYTORCH_RETRY_TEST_CASES: 1
       PYTORCH_OVERRIDE_FLAKY_SIGNAL: 1
       XLA_CLANG_CACHE_S3_BUCKET_NAME: ossci-compiler-clang-cache-circleci-xla
@@ -51,7 +50,7 @@ jobs:
         id: calculate-docker-image
         uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
-          docker-image-name: ${{ DOCKER_IMAGE }}
+          docker-image-name: ${{ env.DOCKER_IMAGE }}
 
       - name: Use following to pull public copy of the image
         id: print-ghcr-mirror
@@ -79,17 +78,11 @@ jobs:
             exit 1
           fi
 
-      - name: Use following to pull public copy of the image
-        id: print-ghcr-mirror
-        shell: bash
-        run: |
-          echo "docker pull ${DOCKER_IMAGE}"
-          docker pull ${DOCKER_IMAGE}
-
       - name: Test
         id: test
         env:
-          BUILD_ENVIRONMENT: ${{ inputs.build-environment }}
+          TEST_COMMAND: .github/scripts/ci_test_xpu.sh
+          DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-py3
           PR_NUMBER: ${{ github.event.pull_request.number }}
           GITHUB_REPOSITORY: ${{ github.repository }}
           GITHUB_WORKFLOW: ${{ github.workflow }}
@@ -107,7 +100,6 @@ jobs:
           # shellcheck disable=SC2086,SC2090
           container_name=$(docker run \
             ${GPU_FLAG:-} \
-            -e BUILD_ENVIRONMENT \
             -e PR_NUMBER \
             -e GITHUB_ACTIONS \
             -e GITHUB_REPOSITORY \
@@ -137,7 +129,7 @@ jobs:
           # save container name for later step
           echo "CONTAINER_NAME=${container_name}" >> "$GITHUB_ENV"
           # jenkins user does not have write permission to mounted workspace; work-around by copying within container to jenkins home
-          docker exec -t "${container_name}" sh -c "bash ${TEST_COMMAND}"
+          docker exec -t "${container_name}" sh -c "bash ${env.TEST_COMMAND}"
 
       - name: Change permissions
         if: ${{ always() && steps.test.conclusion }}

From bae7000dafec4e8204c023560a3c7045d1ace50d Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Mon, 8 Sep 2025 01:16:56 -0700
Subject: [PATCH 17/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index d4ab4e43d5..0e3dd47a6b 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -158,4 +158,4 @@ jobs:
           path: ./**/core.[1-9]*
 
       - name: Teardown XPU
-        uses: pytorch/pytorch/.github/actions/teardown-xpu
+        uses: pytorch/pytorch/.github/actions/teardown-xpu@main

From 7bd3d291d5416e1e976f99c35c76f90268b80db2 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Mon, 8 Sep 2025 23:16:06 -0700
Subject: [PATCH 18/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 65 ++++++++++++++++++++++++++++++-
 1 file changed, 63 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 0e3dd47a6b..4b46f14218 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -29,11 +29,72 @@ jobs:
       XLA_CLANG_CACHE_S3_BUCKET_NAME: ossci-compiler-clang-cache-circleci-xla
     steps:
       # [see note: pytorch repo ref]
+      - name: Checkout PyTorch
+        uses: pytorch/pytorch/.github/actions/checkout-pytorch@main
+
       - name: Checkout Torchao
         uses: actions/checkout@v4
 
-      - name: Setup XPU
-        uses: pytorch/pytorch/.github/actions/setup-xpu@main
+      - name: Clean all stopped docker containers
+        if: always()
+        shell: bash
+        run: |
+          # Prune all stopped containers.
+          # If other runner is pruning on this node, will skip.
+          nprune=$(ps -ef | grep -c "docker container prune")
+          if [[ $nprune -eq 1 ]]; then
+            docker container prune -f
+          fi
+
+      - name: Runner health check system info
+        if: always()
+        shell: bash
+        run: |
+          cat /etc/os-release || true
+          cat /etc/apt/sources.list.d/oneAPI.list || true
+          cat /etc/apt/sources.list.d/intel-gpu-jammy.list || true
+          whoami
+
+      - name: Runner health check xpu-smi
+        if: always()
+        shell: bash
+        run: |
+          timeout 30 xpu-smi discovery || true
+
+      - name: Runner health check GPU count
+        if: always()
+        shell: bash
+        run: |
+          ngpu=$(timeout 30 xpu-smi discovery | grep -c -E 'Device Name' || true)
+          msg="Please file an issue on pytorch/pytorch reporting the faulty runner. Include a link to the runner logs so the runner can be identified"
+          if [[ $ngpu -eq 0 ]]; then
+            echo "Error: Failed to detect any GPUs on the runner"
+            echo "$msg"
+            exit 1
+          fi
+
+      - name: Runner diskspace health check
+        uses: pytorch/pytorch/github/actions/diskspace-cleanup
+        if: always()
+
+      - name: Runner health check disconnect on failure
+        if: ${{ failure() }}
+        shell: bash
+        run: |
+          killall runsvc.sh
+
+      - name: Preserve github env variables for use in docker
+        shell: bash
+        run: |
+          env | grep '^GITHUB' >> "/tmp/github_env_${GITHUB_RUN_ID}"
+          env | grep '^CI' >> "/tmp/github_env_${GITHUB_RUN_ID}"
+
+      - name: XPU set GPU_FLAG
+        shell: bash
+        run: |
+          # Add render group for container creation.
+          render_gid=`cat /etc/group | grep render | cut -d: -f3`
+          echo "GPU_FLAG=--device=/dev/mem --device=/dev/dri --group-add video --group-add $render_gid" >> "${GITHUB_ENV}"
 
       - name: configure aws credentials
         id: aws_creds

From 4fd2909602b09a4d6dbcc83815cbe9b76d03f8a1 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Mon, 8 Sep 2025 23:18:29 -0700
Subject: [PATCH 19/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 4b46f14218..44826d258c 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -74,7 +74,7 @@ jobs:
           fi
 
       - name: Runner diskspace health check
-        uses: pytorch/pytorch/github/actions/diskspace-cleanup
+        uses: pytorch/pytorch/.github/actions/diskspace-cleanup@main
         if: always()
 
       - name: Runner health check disconnect on failure

From 4a7d9af91d38a6911699ea9b0ac654c191e6c758 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Tue, 9 Sep 2025 00:44:07 -0700
Subject: [PATCH 20/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 44826d258c..9d610cbd76 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -195,7 +195,7 @@ jobs:
       - name: Change permissions
         if: ${{ always() && steps.test.conclusion }}
         run: |
-          docker exec -t "${{ env.CONTAINER_NAME }}" sh -c "sudo chown -R 1001:1001 test"
+          docker exec -t "${{ env.CONTAINER_NAME }}" sh -c "sudo chown -R jenkins:jenkins test"
 
       - name: Collect backtraces from coredumps (if any)
         if: always()

From c3f43848c7a3259a8759645b8356d498477391e8 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Tue, 9 Sep 2025 00:47:32 -0700
Subject: [PATCH 21/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 9d610cbd76..12b0726717 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -21,7 +21,7 @@ jobs:
     # Don't run on forked repos or empty test matrix
     # if: github.repository_owner == 'pytorch' && toJSON(fromJSON(inputs.test-matrix).include) != '[]'
     timeout-minutes: 60
-    runs-on: ubuntu-latest
+    runs-on: linux.idc.xpu
     env:
       DOCKER_IMAGE: ci-image:pytorch-linux-jammy-xpu-n-py3
       PYTORCH_RETRY_TEST_CASES: 1

From e8936cb6f7c73b69bdf8136423313d4bcf814035 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Tue, 9 Sep 2025 22:42:45 -0700
Subject: [PATCH 22/25] format fix

---
 .github/workflows/pr-test-xpu.yml | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 12b0726717..07ef696be5 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -12,6 +12,10 @@ on:
     branches:
       - main
 
+permissions:
+  id-token: write
+  contents: read
+
 concurrency:
   group: xpu_ci_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}
   cancel-in-progress: true

From 50e56ec0f8ae28ab130f5761206439469148e30b Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Sun, 14 Sep 2025 23:19:14 -0700
Subject: [PATCH 23/25] trigger by tag only

---
 .github/workflows/pr-test-xpu.yml | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 07ef696be5..6581349206 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -8,9 +8,6 @@ on:
   push:
     tags:
       - ciflow/xpu/*
-  pull_request:
-    branches:
-      - main
 
 permissions:
   id-token: write

From 5a4634112bdce70752211b40d2435d1114d520b5 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Tue, 16 Sep 2025 01:31:17 -0700
Subject: [PATCH 24/25] add xpu label for xpuci

---
 .github/pytorch-probot.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/pytorch-probot.yml b/.github/pytorch-probot.yml
index f0230a8ecd..85ecfdbed2 100644
--- a/.github/pytorch-probot.yml
+++ b/.github/pytorch-probot.yml
@@ -4,3 +4,4 @@ ciflow_push_tags:
 - ciflow/tutorials
 - ciflow/rocm
 - ciflow/4xh100
+- ciflow/xpu

From 030121f128a5a0007b79ffb8998520548873bb67 Mon Sep 17 00:00:00 2001
From: diwei sun <diwei.sun@intel.com>
Date: Tue, 16 Sep 2025 19:29:47 -0700
Subject: [PATCH 25/25] fix docker path

---
 .github/workflows/pr-test-xpu.yml | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.github/workflows/pr-test-xpu.yml b/.github/workflows/pr-test-xpu.yml
index 6581349206..71b28ed76d 100644
--- a/.github/workflows/pr-test-xpu.yml
+++ b/.github/workflows/pr-test-xpu.yml
@@ -113,6 +113,7 @@ jobs:
         uses: pytorch/test-infra/.github/actions/calculate-docker-image@main
         with:
           docker-image-name: ${{ env.DOCKER_IMAGE }}
+          docker-build-dir: pytorch/pytorch/.ci/docker
 
       - name: Use following to pull public copy of the image
         id: print-ghcr-mirror