Merge branch 'main' into jz/re-land-ao-api

Author: jackzhxng

.ci/scripts/build_android_instrumentation.sh

@@ -12,30 +12,10 @@ if [[ -z "${PYTHON_EXECUTABLE:-}" ]]; then
 fi
 which "${PYTHON_EXECUTABLE}"
 
-build_android_test() {
-  mkdir -p extension/android/executorch_android/src/androidTest/resources
-  cp extension/module/test/resources/add.pte extension/android/executorch_android/src/androidTest/resources
-  pushd extension/android
-  ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew :executorch_android:testDebugUnitTest
-  ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew :executorch_android:assembleAndroidTest
-  popd
-}
+mkdir -p "${BUILD_AAR_DIR}"/executorch_android/src/androidTest/resources
+cp extension/module/test/resources/add.pte "${BUILD_AAR_DIR}"/executorch_android/src/androidTest/resources
 
-collect_artifacts_to_be_uploaded() {
-  ARTIFACTS_DIR_NAME="$1"
-  # Collect Java library test
-  JAVA_LIBRARY_TEST_DIR="${ARTIFACTS_DIR_NAME}/library_test_dir"
-  mkdir -p "${JAVA_LIBRARY_TEST_DIR}"
-  cp extension/android/executorch_android/build/outputs/apk/androidTest/debug/*.apk "${JAVA_LIBRARY_TEST_DIR}"
-}
-
-main() {
-  build_android_test
-  if [ -n "$ARTIFACTS_DIR_NAME" ]; then
-    collect_artifacts_to_be_uploaded ${ARTIFACTS_DIR_NAME}
-  fi
-}
-
-if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
-  main "$@"
-fi
+pushd "${BUILD_AAR_DIR}"
+ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew :executorch_android:testDebugUnitTest
+ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew :executorch_android:assembleAndroidTest
+popd

.ci/scripts/unittest-macos.sh

@@ -19,14 +19,14 @@ export TMP_DIR=$(mktemp -d)
 export PATH="${TMP_DIR}:$PATH"
 trap 'rm -rfv ${TMP_DIR}' EXIT
 
-if [[ "$BUILD_TOOL" == "cmake" ]]; then
-    # Setup MacOS dependencies as there is no Docker support on MacOS atm
-    PYTHON_EXECUTABLE=python \
-    EXECUTORCH_BUILD_PYBIND=ON \
-    CMAKE_ARGS="-DEXECUTORCH_BUILD_COREML=ON -DEXECUTORCH_BUILD_MPS=ON -DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON" \
-    ${CONDA_RUN} --no-capture-output \
-    .ci/scripts/setup-macos.sh "$@"
+# Setup MacOS dependencies as there is no Docker support on MacOS atm
+PYTHON_EXECUTABLE=python \
+EXECUTORCH_BUILD_PYBIND=ON \
+CMAKE_ARGS="-DEXECUTORCH_BUILD_COREML=ON -DEXECUTORCH_BUILD_MPS=ON -DEXECUTORCH_BUILD_XNNPACK=ON -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON" \
+${CONDA_RUN} --no-capture-output \
+.ci/scripts/setup-macos.sh "$@"
 
+if [[ "$BUILD_TOOL" == "cmake" ]]; then
     # Install llama3_2_vision dependencies.
     PYTHON_EXECUTABLE=python \
     ${CONDA_RUN} --no-capture-output \

.ci/scripts/utils.sh

@@ -60,12 +60,46 @@ install_pytorch_and_domains() {
   # Fetch the target commit
   pushd pytorch || return
   git checkout "${TORCH_VERSION}"
-  git submodule update --init --recursive
 
-  export USE_DISTRIBUTED=1
-  # Then build and install PyTorch
-  python setup.py bdist_wheel
-  pip install "$(echo dist/*.whl)"
+  local system_name=$(uname)
+  if [[ "${system_name}" == "Darwin" ]]; then
+    local platform=$(python -c 'import sysconfig; import platform; v=platform.mac_ver()[0].split(".")[0]; platform=sysconfig.get_platform().split("-"); platform[1]=f"{v}_0"; print("_".join(platform))')
+  fi
+  local python_version=$(python -c 'import platform; v=platform.python_version_tuple(); print(f"{v[0]}{v[1]}")')
+  local torch_release=$(cat version.txt)
+  local torch_short_hash=${TORCH_VERSION:0:7}
+  local torch_wheel_path="cached_artifacts/pytorch/executorch/pytorch_wheels/${system_name}/${python_version}"
+  local torch_wheel_name="torch-${torch_release}%2Bgit${torch_short_hash}-cp${python_version}-cp${python_version}-${platform:-}.whl"
+
+  local cached_torch_wheel="https://gha-artifacts.s3.us-east-1.amazonaws.com/${torch_wheel_path}/${torch_wheel_name}"
+  # Cache PyTorch wheel is only needed on MacOS, Linux CI already has this as part
+  # of the Docker image
+  local torch_wheel_not_found=0
+  if [[ "${system_name}" == "Darwin" ]]; then
+    pip install "${cached_torch_wheel}" || torch_wheel_not_found=1
+  else
+    torch_wheel_not_found=1
+  fi
+
+  # Found no such wheel, we will build it from source then
+  if [[ "${torch_wheel_not_found}" == "1" ]]; then
+    echo "No cached wheel found, continue with building PyTorch at ${TORCH_VERSION}"
+
+    git submodule update --init --recursive
+    USE_DISTRIBUTED=1 python setup.py bdist_wheel
+    pip install "$(echo dist/*.whl)"
+
+    # Only AWS runners have access to S3
+    if command -v aws && [[ -z "${GITHUB_RUNNER:-}" ]]; then
+      for wheel_path in dist/*.whl; do
+        local wheel_name=$(basename "${wheel_path}")
+        echo "Caching ${wheel_name}"
+        aws s3 cp "${wheel_path}" "s3://gha-artifacts/${torch_wheel_path}/${wheel_name}"
+      done
+    fi
+  else
+    echo "Use cached wheel at ${cached_torch_wheel}"
+  fi
 
   # Grab the pinned audio and vision commits from PyTorch
   TORCHAUDIO_VERSION=$(cat .github/ci_commit_pins/audio.txt)

.github/workflows/_android.yml

@@ -27,16 +27,21 @@ jobs:
         conda activate "${CONDA_ENV}"
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh --build-tool buck2
         export ARTIFACTS_DIR_NAME=artifacts-to-be-uploaded
-
-        mkdir -p ${ARTIFACTS_DIR_NAME}/fp32-xnnpack-custom
-        bash examples/models/llama/install_requirements.sh
-        bash ".ci/scripts/test_llama.sh" -model stories110M -build_tool cmake -dtype fp16 -mode portable -upload ${ARTIFACTS_DIR_NAME}/fp32-xnnpack-custom
+        mkdir -p ${ARTIFACTS_DIR_NAME}/
 
         # Build LLM Demo for Android
         export BUILD_AAR_DIR=aar-out
         mkdir -p $BUILD_AAR_DIR
-        bash build/build_android_library.sh ${ARTIFACTS_DIR_NAME}
-        bash .ci/scripts/build_android_instrumentation.sh ${ARTIFACTS_DIR_NAME}
+        bash scripts/build_android_library.sh
+        cp ${BUILD_AAR_DIR}/executorch.aar $ARTIFACTS_DIR_NAME
+
+        mkdir -p ${ARTIFACTS_DIR_NAME}/library_test_dir
+        bash .ci/scripts/build_android_instrumentation.sh
+        cp ${BUILD_AAR_DIR}/executorch_android/build/outputs/apk/androidTest/debug/executorch_android-debug-androidTest.apk "${ARTIFACTS_DIR_NAME}/library_test_dir"
+
+        mkdir -p ${ARTIFACTS_DIR_NAME}/fp32-xnnpack-custom
+        bash examples/models/llama/install_requirements.sh
+        bash ".ci/scripts/test_llama.sh" -model stories110M -build_tool cmake -dtype fp16 -mode portable -upload ${ARTIFACTS_DIR_NAME}/fp32-xnnpack-custom
 
         mkdir -p examples/demo-apps/android/LlamaDemo/app/libs
         cp aar-out/executorch.aar examples/demo-apps/android/LlamaDemo/app/libs
@@ -96,7 +101,7 @@ jobs:
           curl -O https://gha-artifacts.s3.amazonaws.com/${{ github.repository }}/${{ github.run_id }}/artifacts/llm_demo/app-debug.apk
           curl -O https://gha-artifacts.s3.amazonaws.com/${{ github.repository }}/${{ github.run_id }}/artifacts/llm_demo/app-debug-androidTest.apk
           curl -O https://gha-artifacts.s3.amazonaws.com/${{ github.repository }}/${{ github.run_id }}/artifacts/fp32-xnnpack-custom/model.zip
-          curl -o android-test-debug-androidTest.apk https://gha-artifacts.s3.amazonaws.com/${{ github.repository }}/${{ github.run_id }}/artifacts/library_test_dir/executorch-debug-androidTest.apk
+          curl -o android-test-debug-androidTest.apk https://gha-artifacts.s3.amazonaws.com/${{ github.repository }}/${{ github.run_id }}/artifacts/library_test_dir/executorch_android-debug-androidTest.apk
           unzip model.zip
           mv *.pte model.pte
 
@@ -120,7 +125,7 @@ jobs:
         with:
           api-level: ${{ env.API_LEVEL }}
           arch: x86_64
-          script: ./build/run_android_emulator.sh
+          script: ./scripts/run_android_emulator.sh
           # NB: This is to boot the emulator faster following the instructions on
           # https://github.com/ReactiveCircus/android-emulator-runner. The max number
           # of cores we can set is 6, any higher number will be reduced to 6.

.github/workflows/_unittest.yml

@@ -49,4 +49,6 @@ jobs:
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
         set -eux
+        # This is needed to get the prebuilt PyTorch wheel from S3
+        ${CONDA_RUN} --no-capture-output pip install awscli==1.37.21
         .ci/scripts/unittest-macos.sh --build-tool "${{ inputs.build-tool }}" --build-mode "${{ inputs.build-mode }}" --editable "${{ inputs.editable }}"

.github/workflows/android-perf.yml

@@ -363,7 +363,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/build-qnn-sdk.sh
 
         mkdir -p aar-out
-        PYTHON_EXECUTABLE=python ANDROID_ABIS="arm64-v8a" BUILD_AAR_DIR=aar-out EXECUTORCH_BUILD_QNN=ON QNN_SDK_ROOT=/tmp/qnn/2.28.0.241029 bash build/build_android_library.sh
+        PYTHON_EXECUTABLE=python ANDROID_ABIS="arm64-v8a" BUILD_AAR_DIR=aar-out EXECUTORCH_BUILD_QNN=ON QNN_SDK_ROOT=/tmp/qnn/2.28.0.241029 bash scripts/build_android_library.sh
         mkdir -p extension/benchmark/android/benchmark/app/libs
         cp aar-out/executorch.aar extension/benchmark/android/benchmark/app/libs
         pushd extension/benchmark/android/benchmark

.github/workflows/android-release-artifacts.yml

@@ -55,7 +55,7 @@ jobs:
         # Build AAR Package
         mkdir aar-out
         export BUILD_AAR_DIR=aar-out
-        bash build/build_android_library.sh
+        bash scripts/build_android_library.sh
         mkdir -p "${ARTIFACTS_DIR_NAME}"
         cp aar-out/executorch.aar "${ARTIFACTS_DIR_NAME}/executorch.aar"
 

.github/workflows/apple-perf.yml

@@ -387,7 +387,7 @@ jobs:
 
         echo "::group::Build ExecuTorch iOS frameworks"
         PYTHON_EXECUTABLE=python ${CONDA_RUN} --no-capture-output \
-          build/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
+          scripts/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
         echo "::endgroup::"
 
         # NB: Although exported models can be copied to this directory and bundled together with the
@@ -396,7 +396,7 @@ jobs:
         # create the directory here to pass the build
         mkdir -p extension/benchmark/apple/Benchmark/Models
         ${CONDA_RUN} --no-capture-output \
-          build/build_apple_llm_demo.sh ${ARTIFACTS_DIR_NAME}
+          scripts/build_apple_llm_demo.sh ${ARTIFACTS_DIR_NAME}
 
   upload-benchmark-app:
     needs: build-benchmark-app

.github/workflows/apple.yml

@@ -11,9 +11,9 @@ on:
       - .github/workflows/apple.yml
       - install_executorch.sh
       - backends/apple/**
-      - build/build_apple_frameworks.sh
-      - build/build_apple_llm_demo.sh
-      - build/create_frameworks.sh
+      - scripts/build_apple_frameworks.sh
+      - scripts/build_apple_llm_demo.sh
+      - scripts/create_frameworks.sh
       - .ci/scripts/test_ios_ci.sh
       - examples/demo-apps/apple_ios/**
       - extension/apple/**
@@ -172,7 +172,7 @@ jobs:
 
         # Build iOS Frameworks
         PYTHON_EXECUTABLE=python ${CONDA_RUN} --no-capture-output \
-        build/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
+        scripts/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
 
         # Bundle iOS Frameworks
         for FRAMEWORK in "${FRAMEWORKS[@]}"; do (
@@ -313,11 +313,11 @@ jobs:
 
         echo "::group::Build ExecuTorch iOS frameworks"
         PYTHON_EXECUTABLE=python ${CONDA_RUN} --no-capture-output \
-          build/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
+          scripts/build_apple_frameworks.sh --Release --Debug --coreml --custom --mps --optimized --portable --quantized --xnnpack
         echo "::endgroup::"
 
         echo "::group::Build ExecuTorch benchmark app"
         mkdir -p extension/benchmark/apple/Benchmark/Models
         ${CONDA_RUN} --no-capture-output \
-          build/build_apple_llm_demo.sh "${ARTIFACTS_DIR_NAME}"
+          scripts/build_apple_llm_demo.sh "${ARTIFACTS_DIR_NAME}"
         echo "::endgroup::"

.github/workflows/pull.yml

@@ -290,6 +290,36 @@ jobs:
         # run e2e (export, tokenizer and runner)
         PYTHON_EXECUTABLE=python bash .ci/scripts/test_llava.sh
 
+  test-moshi-linux:
+    name: test-moshi-linux
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
+    strategy:
+      fail-fast: false
+    with:
+      runner: linux.2xlarge
+      docker-image: executorch-ubuntu-22.04-clang12
+      submodules: 'true'
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 90
+      script: |
+        # The generic Linux job chooses to use base env, not the one setup by the image
+        CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
+        conda activate "${CONDA_ENV}"
+
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh --build-tool "cmake"
+
+        # install Mimi requirements
+        bash examples/models/moshi/mimi/install_requirements.sh
+
+        # reinstall executorch
+        bash ./install_executorch.sh
+
+        # run python unittest
+        python -m unittest examples.models.moshi.mimi.test_mimi
+
   test-quantized-aot-lib-linux:
     name: test-quantized-aot-lib-linux
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main

.github/workflows/trunk.yml

@@ -228,7 +228,7 @@ jobs:
     name: test-coreml-delegate
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     with:
-      runner: macos-13-xlarge
+      runner: macos-latest-xlarge
       python-version: '3.11'
       submodules: 'true'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}

.gitignore

@@ -11,6 +11,7 @@ __pycache__/
 arm_test/
 buck-out/
 buck2-bin/
+build/
 cmake-android-out/
 cmake-ios-out/
 cmake-out*

README.md

@@ -67,7 +67,6 @@ We welcome contributions. To get started review the [guidelines](CONTRIBUTING.md
 ```
 executorch
 ├── backends                        #  Backend delegate implementations.
-├── build                           #  Utilities for managing the build system.
 ├── codegen                         #  Tooling to autogenerate bindings between kernels and the runtime.
 ├── configurations
 ├── docs                            #  Static docs tooling.
@@ -81,8 +80,7 @@ executorch
 |   ├── operator                    #  Operator node manipulation utilities.
 |   ├── passes                      #  Built-in compiler passes.
 |   ├── program                     #  Export artifacts.
-|   ├── serde                       #  Graph module
-serialization/deserialization.
+|   ├── serde                       #  Graph module serialization/deserialization.
 |   ├── verification                #  IR verification.
 ├── extension                       #  Extensions built on top of the runtime.
 |   ├── android                     #  ExecuTorch wrappers for Android apps.
@@ -97,8 +95,7 @@ serialization/deserialization.
 |   ├── parallel                    #  C++ threadpool integration.
 |   ├── pybindings                  #  Python API for executorch runtime.
 |   ├── pytree                      #  C++ and Python flattening and unflattening lib for pytrees.
-|   ├── runner_util                 #  Helpers for writing C++ PTE-execution
-tools.
+|   ├── runner_util                 #  Helpers for writing C++ PTE-execution tools.
 |   ├── testing_util                #  Helpers for writing C++ tests.
 |   ├── training                    #  Experimental libraries for on-device training
 ├── kernels                         #  1st party kernel implementations.
@@ -114,9 +111,9 @@ tools.
 |   ├── executor                    #  Model loading, initialization, and execution.
 |   ├── kernel                      #  Kernel registration and management.
 |   ├── platform                    #  Layer between architecture specific code and portable C++.
-├── schema                          #  ExecuTorch PTE file format flatbuffer
-schemas.
-├── scripts                         #  Utility scripts for size management, dependency management, etc.
+├── schema                          #  ExecuTorch PTE file format flatbuffer schemas.
+├── scripts                         #  Utility scripts for building libs, size management, dependency management, etc.
+├── tools                           #  Development tool management.
 ├── devtools                        #  Model profiling, debugging, and introspection.
 ├── shim                            #  Compatibility layer between OSS and Internal builds
 ├── test                            #  Broad scoped end-to-end tests.

backends/apple/coreml/setup.md

@@ -50,7 +50,7 @@ xcode-select --install
 
 ```bash
 cd executorch
-./build/build_apple_frameworks.sh --coreml
+./scripts/build_apple_frameworks.sh --coreml
 ```
 5. Open the project in Xcode, and drag `executorch.xcframework` and `coreml_backend.xcframework` frameworks generated from Step 2 to Frameworks.
 

backends/apple/mps/setup.md

@@ -130,7 +130,7 @@ python3 -m sdk.inspector.inspector_cli --etdump_path etdump.etdp --etrecord_path
 ***Step 1***. Create the ExecuTorch core and MPS delegate frameworks to link on iOS
 ```bash
 cd executorch
-./build/build_apple_frameworks.sh --mps
+./scripts/build_apple_frameworks.sh --mps
 ```
 
 `mps_delegate.xcframework` will be in `cmake-out` folder, along with `executorch.xcframework` and `portable_delegate.xcframework`:

backends/arm/_passes/insert_table_ops.py

@@ -41,6 +41,7 @@ class InsertTableOpsPass(ExportPass):
     """
 
     table_ops: Dict[EdgeOpOverload, Callable[[torch.Tensor], torch.Tensor]] = {
+        exir_ops.edge.aten.ceil.default: torch.ceil,
         exir_ops.edge.aten.exp.default: torch.exp,
         exir_ops.edge.aten.floor.default: torch.floor,
         exir_ops.edge.aten.log.default: torch.log,