Update on "Use extension module for pybind"

Deprecate the module class in pybindings, previously it was blocked by bundled program, and now it's resolved with https://github.com/pytorch/executorch/blob/2bb567f15a54fbc0ef621e5e85ff24e5da505b11/extension/module/bundled_module.h#L4

Differential Revision: [D70516347](https://our.internmc.facebook.com/intern/diff/D70516347/)

[ghstack-poisoned]

Author: lucylq

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

@@ -16,6 +16,7 @@ on:
         options:
           - "xnnpack"
           - "vulkan+xnnpack"
+          - "qnn"
   schedule:
     - cron: 0 10 * * *
 
@@ -83,17 +84,29 @@ jobs:
 
         echo -n "$SECRET_EXECUTORCH_MAVEN_SIGNING_GPG_KEY_CONTENTS" | base64 -d > /tmp/secring.gpg
 
+        GRADLE_ARGS=""
+
         # Update the version name in build.gradle in case of maven publish
         VERSION="${{ inputs.version }}"
         if [ ! -z "$VERSION" ]; then
-          sed -i "s/\(coordinates(\"org.pytorch\", \"executorch-android\", \"\)\([0-9]\+.[0-9]\+.[0-9]\+\)\(\")\)/\1$VERSION\3/" extension/android/executorch_android/build.gradle
+          GRADLE_ARGS+=" -DexecuTorchVersion=${VERSION}"
         fi
 
         FLAVOR="${{ inputs.flavor }}"
         if [[ "$FLAVOR" == "vulkan+xnnpack" || -z "$FLAVOR" ]]; then
           export EXECUTORCH_BUILD_VULKAN=ON
         fi
 
+        if [[ "$FLAVOR" == "qnn" ]]; then
+          PYTHON_EXECUTABLE=python bash .ci/scripts/setup-qnn-deps.sh
+          PYTHON_EXECUTABLE=python bash .ci/scripts/build-qnn-sdk.sh
+          export EXECUTORCH_BUILD_QNN=ON
+          source backends/qualcomm/scripts/qnn_config.sh
+          export QNN_SDK_ROOT="/tmp/qnn/${QNN_VERSION}"
+          export ANDROID_ABIS=arm64-v8a
+          GRADLE_ARGS+=" -DqnnVersion=2.28.0"
+        fi
+
         # Build AAR Package
         mkdir aar-out
         export BUILD_AAR_DIR=aar-out
@@ -106,7 +119,7 @@ jobs:
         # Publish to maven staging
         UPLOAD_TO_MAVEN="${{ inputs.upload_to_maven }}"
         if [[ "$UPLOAD_TO_MAVEN" == "true" ]]; then
-          (cd extension/android; ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew :executorch_android:publishToMavenCentral)
+          (cd extension/android; ANDROID_HOME="${ANDROID_SDK:-/opt/android/sdk}" ./gradlew  ${GRADLE_ARGS} :executorch_android:publishToMavenCentral)
         fi
 
   upload-release-aar:

.github/workflows/apple-perf.yml

@@ -416,7 +416,7 @@ jobs:
       - set-parameters
     secrets: inherit
     with:
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: '3.11'
       submodules: 'recursive'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}

.github/workflows/apple.yml

@@ -49,7 +49,7 @@ jobs:
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     secrets: inherit
     with:
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: '3.11'
       submodules: 'recursive'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -136,7 +136,7 @@ jobs:
     needs: set-version
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     with:
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: '3.11'
       submodules: 'recursive'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -276,7 +276,7 @@ jobs:
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     secrets: inherit
     with:
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: '3.11'
       submodules: 'recursive'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}

.github/workflows/build-presets.yml

@@ -22,7 +22,7 @@ jobs:
     with:
       job-name: build
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: 3.12
       submodules: recursive
       timeout: 90
@@ -109,7 +109,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        preset: [windows]
+        preset: [pybind, windows]
     with:
       job-name: build
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -122,14 +122,17 @@ jobs:
           Set-PSDebug -Trace 1
           \$ErrorActionPreference = 'Stop'
           \$PSNativeCommandUseErrorActionPreference = \$true
+
           conda create --yes --quiet -n et python=3.12
           conda activate et
           python install_requirements.py
+
           cmake --preset ${{ matrix.preset }} -T ClangCL
           if (\$LASTEXITCODE -ne 0) {
             Write-Host "CMake configuration was unsuccessful. Exit code: \$LASTEXITCODE."
             exit \$LASTEXITCODE
           }
+
           \$numCores = [System.Environment]::GetEnvironmentVariable('NUMBER_OF_PROCESSORS') - 1
           cmake --build cmake-out -j \$numCores
           if (\$LASTEXITCODE -ne 0) {

.github/workflows/build-wheels-macos.yml

@@ -63,7 +63,7 @@ jobs:
       post-script: ${{ matrix.post-script }}
       package-name: ${{ matrix.package-name }}
       # Meta's macOS runners do not have Xcode, so use GitHub's runners.
-      runner-type: macos-latest-xlarge
+      runner-type: macos-14-xlarge
       setup-miniconda: true
       smoke-test-script: ${{ matrix.smoke-test-script }}
       trigger-event: ${{ github.event_name }}

.github/workflows/pull.yml

@@ -855,7 +855,8 @@ jobs:
         .ci/scripts/setup-linux.sh --build-tool "cmake"
 
         # Install test requirements
-        pip install -r backends/nxp/requirements-tests.txt
+        pip install -r backends/nxp/requirements-tests-pypi.txt
+        pip install -r backends/nxp/requirements-tests-eiq.txt
 
         # Run pytest
         PYTHON_EXECUTABLE=python bash backends/nxp/run_unittests.sh

.github/workflows/trunk.yml

@@ -435,7 +435,7 @@ jobs:
     name: test-coreml-delegate
     uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
     with:
-      runner: macos-latest-xlarge
+      runner: macos-14-xlarge
       python-version: '3.11'
       submodules: 'recursive'
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -940,3 +940,42 @@ jobs:
       build-mode: Release
       build-tool: cmake
       docker-image: ci-image:executorch-ubuntu-22.04-clang12
+
+  test-mcu-models:
+    name: test-mcu-models
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    strategy:
+      matrix:
+        include:
+          - build-tool: cmake
+      fail-fast: false
+    permissions:
+      id-token: write
+      contents: read
+    with:
+      runner: linux.2xlarge
+      docker-image: ci-image:executorch-ubuntu-22.04-arm-sdk
+      submodules: 'recursive'
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 90
+      script: |
+        BUILD_TOOL=${{ matrix.build-tool }}
+
+        # 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}"
+
+        # Try to mirror these as closely as possible
+        source .ci/scripts/utils.sh
+        install_executorch "--use-pt-pinned-commit"
+
+        .ci/scripts/setup-arm-baremetal-tools.sh
+        source examples/arm/ethos-u-scratch/setup_path.sh
+
+        # Run selective Build
+        chmod +x examples/selective_build/test_selective_build.sh
+        examples/selective_build/test_selective_build.sh "${BUILD_TOOL}"
+
+        # Run MCU models
+        chmod +x examples/arm/run_mcu_models_fvp.sh
+        examples/arm/run_mcu_models_fvp.sh --target=cortex-m55

CMakeLists.txt

@@ -582,7 +582,6 @@ endif()
 
 if(EXECUTORCH_BUILD_EXTENSION_APPLE)
   add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/extension/apple)
-  list(APPEND _executorch_extensions apple_extension)
 endif()
 
 if(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER)
@@ -649,7 +648,6 @@ endif()
 
 if(EXECUTORCH_BUILD_EXTENSION_LLM_APPLE)
   add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/extension/llm/apple)
-  list(APPEND _executorch_extensions extension_llm_apple)
 endif()
 
 if(EXECUTORCH_BUILD_EXTENSION_RUNNER_UTIL)
@@ -968,6 +966,10 @@ else()
 endif()
 target_link_libraries(executorch_kernels INTERFACE ${_executorch_kernels})
 
+install(TARGETS executorch_backends executorch_extensions executorch_kernels
+        EXPORT ExecuTorchTargets
+)
+
 if(EXECUTORCH_BUILD_EXECUTOR_RUNNER)
   # Baseline libraries that executor_runner will link against.
   set(_executor_runner_libs executorch extension_evalue_util

backends/apple/coreml/compiler/torch_ops.py

@@ -15,7 +15,6 @@
 from coremltools.converters.mil.frontend.torch.ops import (
     _get_inputs,
     _get_kwinputs,
-    noop,
     NUM_TO_NUMPY_DTYPE,
     NUM_TO_TORCH_DTYPE,
     split,
@@ -92,28 +91,6 @@ def _to_dim_order_copy(context, node):
         to(context, node)
 
 
-@register_torch_op(
-    torch_alias=[
-        "dim_order_ops::_clone_dim_order",
-        "dim_order_ops._clone_dim_order",
-    ],
-    override=False,
-)
-def _clone_dim_order(context, node):
-    dim_order = _get_kwinputs(context, node, "dim_order", default=[None])[0]
-    node.kwinputs.pop("dim_order")
-
-    # In CoreML, dim_order.val will be a ndarray, so we convert it to a list to check memory format.
-    dim_order = [int(d) for d in dim_order.val]
-    memory_format = get_memory_format(dim_order)
-    assert (
-        memory_format == _torch.contiguous_format
-    ), "Only contiguous memory format is supported in CoreML"
-
-    # Since CoreML only supports contiguous format, no dim_order preservation is needed. Treat this as a no-op clone.
-    noop(context, node)
-
-
 # https://github.com/apple/coremltools/pull/2558
 @register_torch_op(
     torch_alias=["torchao::dequantize_affine", "torchao.dequantize_affine"],

backends/apple/coreml/test/test_torch_ops.py

@@ -213,28 +213,6 @@ def test_dequantize_codebook_embedding(self):
         et_prog = delegated_program.to_executorch()
         self._compare_outputs(et_prog, model, example_inputs)
 
-    def test__clone_dim_order_contiguous(self):
-        class Model(torch.nn.Module):
-            def forward(self, x):
-                return torch.ops.dim_order_ops._clone_dim_order(
-                    x, dim_order=[0, 1, 2, 3]
-                )
-
-        model, example_inputs = Model(), (torch.randn(1, 3, 8, 8),)
-        ep = torch.export.export(model, example_inputs)
-        delegated_program = executorch.exir.to_edge_transform_and_lower(
-            ep,
-            partitioner=[self._coreml_partitioner()],
-        )
-        for node in delegated_program.exported_program().graph.nodes:
-            if node.op == "call_function":
-                assert node.target.__name__ in [
-                    "executorch_call_delegate",
-                    "getitem",
-                ], f"Got unexpected node target after delegation: {node.target.__name__}"
-        et_prog = delegated_program.to_executorch()
-        self._compare_outputs(et_prog, model, example_inputs)
-
 
 if __name__ == "__main__":
     test_runner = TestTorchOps()
@@ -245,4 +223,3 @@ def forward(self, x):
     test_runner.test_dequantize_affine_c8w_embedding_b4w_linear()
     test_runner.test_dequantize_codebook_linear()
     test_runner.test_dequantize_codebook_embedding()
-    test_runner.test__clone_dim_order_contiguous()