Merge branch 'pytorch:main' into main

Author: sandeepgupta12

.ci/docker/ci_commit_pins/executorch.txt

@@ -1 +1 @@
-41e7ffa8b7ff09206aa5b9b5c1bbd82b9e0ff277
+319c8d7fd3551bac63429334509de2663aa43f57

.ci/docker/common/install_cuda.sh

@@ -239,7 +239,7 @@ function prune_126 {
 }
 
 function install_128 {
-  CUDNN_VERSION=9.7.0.66
+  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_cuda_aarch64.sh

@@ -161,7 +161,7 @@ function prune_126 {
 }
 
 function install_128 {
-  CUDNN_VERSION=9.7.0.66
+  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

@@ -4,7 +4,9 @@ if [[ -n "${CUDNN_VERSION}" ]]; then
     # cuDNN license: https://developer.nvidia.com/cudnn/license_agreement
     mkdir tmp_cudnn
     pushd tmp_cudnn
-    if [[ ${CUDA_VERSION:0:4} == "12.6" ]]; then
+    if [[ ${CUDA_VERSION:0:4} == "12.8" ]]; then
+        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
         CUDNN_NAME="cudnn-linux-x86_64-9.1.0.70_cuda12-archive"

.ci/docker/common/install_onnx.sh

@@ -39,7 +39,7 @@ pip_install ml_dtypes
 # Cache the transformers model to be used later by ONNX tests. We need to run the transformers
 # package to download the model. By default, the model is cached at ~/.cache/huggingface/hub/
 IMPORT_SCRIPT_FILENAME="/tmp/onnx_import_script.py"
-as_jenkins echo 'import transformers; transformers.AutoModel.from_pretrained("sshleifer/tiny-gpt2"); transformers.AutoTokenizer.from_pretrained("sshleifer/tiny-gpt2"); transformers.AutoModelForSpeechSeq2Seq.from_pretrained("openai/whisper-large-v3");' > "${IMPORT_SCRIPT_FILENAME}"
+as_jenkins echo 'import transformers; transformers.GPTJForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gptj");' > "${IMPORT_SCRIPT_FILENAME}"
 
 # Need a PyTorch version for transformers to work
 pip_install --pre torch --index-url https://download.pytorch.org/whl/nightly/cpu

.ci/pytorch/test.sh

@@ -420,6 +420,7 @@ test_inductor_cpp_wrapper_shard() {
   python test/run_test.py \
     --include inductor/test_torchinductor inductor/test_max_autotune inductor/test_cpu_repro \
     --verbose
+  python test/run_test.py --inductor --include test_torch -k 'take' --verbose
 
   # Run inductor benchmark tests with cpp wrapper.
   # Skip benchmark tests if it's in rerun-disabled-mode.

.github/actions/upload-utilization-stats/action.yml

@@ -0,0 +1,56 @@
+name: upload-utilization-stats
+
+description: Upload utilization stats to artifacts
+
+inputs:
+    workflow_run_id:
+      type: string
+      description: 'workflow (run) id of the workflow the test is running'
+      required: True
+    workflow_attempt:
+      type: string
+      description: 'the workflow (run) attempt'
+      required: True
+    workflow_name:
+      description: 'name of the workflow'
+      type: string
+      required: True
+    job_id:
+      type: string
+      description: 'the job (run) id for the test'
+      required: True
+    job_name:
+      type: string
+      description: 'the job name of the test'
+      required: True
+
+runs:
+  using: composite
+  steps:
+    - name: Print Inputs
+      shell: bash
+      run: |
+        echo "workflow_id: ${{inputs.workflow_run_id}}"
+        echo "workflow_attempt: ${{inputs.workflow_attempt}}"
+        echo "workflow_Name: ${{inputs.workflow_name}}"
+        echo "job_id: ${{inputs.job_id}}"
+        echo "job_name:  ${{inputs.job_name}}"
+    - uses: nick-fields/[email protected]
+      name: Setup dependencies
+      with:
+        shell: bash
+        timeout_minutes: 5
+        max_attempts: 5
+        retry_wait_seconds: 30
+        command: |
+          set -eu
+          python3 -m pip install python-dateutil==2.8.2 boto3==1.35.42 pandas==2.1.3
+    - name: Upload utilizatoin stats to s3
+      shell: bash
+      run: |
+        python3 -m tools.stats.upload_utilization_stats.upload_utilization_stats \
+          --workflow-run-id "${{inputs.workflow_run_id}}" \
+          --workflow-name "${{inputs.workflow_name}}" \
+          --workflow-run-attempt "${{inputs.workflow_attempt}}" \
+          --job-id "${{inputs.job_id}}" \
+          --job-name "${{inputs.job_name}}"

.github/ci_commit_pins/audio.txt

@@ -1 +1 @@
-2709b65c9d3c55da40a5436ec4c45c427feb1d2a
+f084f34bbb743fada85f66b0ed8041387565e69c

.github/scripts/generate_binary_build_matrix.py

@@ -93,7 +93,7 @@
         "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==9.7.0.66; 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' | "

.github/workflows/_linux-test.yml

@@ -416,6 +416,17 @@ jobs:
           if-no-files-found: ignore
           path: ./**/core.[1-9]*
 
+      - name: Upload utilization stats
+        if: ${{ always() && steps.test.conclusion && steps.test.conclusion != 'skipped' && !inputs.disable-monitor }}
+        continue-on-error: true
+        uses: ./.github/actions/upload-utilization-stats
+        with:
+          job_id: ${{ steps.get-job-id.outputs.job-id }}
+          job_name: ${{ steps.get-job-id.outputs.job-name }}
+          workflow_name: ${{ github.workflow }}
+          workflow_run_id: ${{github.run_id}}
+          workflow_attempt: ${{github.run_attempt}}
+
       - name: Teardown Linux
         uses: pytorch/test-infra/.github/actions/teardown-linux@main
         if: always() && steps.check_container_runner.outputs.IN_CONTAINER_RUNNER == 'false'