Update on "[xnnpack] Refactor test/op_linear"

No functional change, just moving code around.

Differential Revision: [D68350953](https://our.internmc.facebook.com/intern/diff/D68350953/)

[ghstack-poisoned]

Author: digantdesai

.ci/scripts/utils.sh

@@ -17,17 +17,17 @@ retry () {
 }
 
 clean_executorch_install_folders() {
-  ./install_requirements.sh --clean
+  ./install_executorch.sh --clean
 }
 
 install_executorch() {
   which pip
   # Install executorch, this assumes that Executorch is checked out in the
   # current directory.
   if [[ "${1:-}" == "use-pt-pinned-commit" ]]; then
-    ./install_requirements.sh --pybind xnnpack --use-pt-pinned-commit
+    ./install_executorch.sh --pybind xnnpack --use-pt-pinned-commit
   else
-    ./install_requirements.sh --pybind xnnpack
+    ./install_executorch.sh --pybind xnnpack
   fi
   # Just print out the list of packages for debugging
   pip list

.github/workflows/apple.yml

@@ -9,7 +9,7 @@ on:
     paths:
       - .ci/scripts/setup-ios.sh
       - .github/workflows/apple.yml
-      - install_requirements.sh
+      - install_executorch.sh
       - backends/apple/**
       - build/build_apple_frameworks.sh
       - build/build_apple_llm_demo.sh

.github/workflows/pull.yml

@@ -200,7 +200,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install Llava requirements
         bash examples/models/llama/install_requirements.sh
@@ -333,6 +333,9 @@ jobs:
 
   unittest-arm:
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk
@@ -433,7 +436,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install phi-3-mini requirements
         bash examples/models/phi-3-mini/install_requirements.sh
@@ -460,7 +463,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh
@@ -487,7 +490,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh
@@ -514,7 +517,7 @@ jobs:
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh "cmake"
 
         # install pybind
-        bash install_requirements.sh --pybind xnnpack
+        bash install_executorch.sh --pybind xnnpack
 
         # install llama requirements
         bash examples/models/llama/install_requirements.sh

.github/workflows/trunk.yml

@@ -132,6 +132,9 @@ jobs:
   test-arm-backend-delegation:
     name: test-arm-backend-delegation
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk
@@ -159,6 +162,9 @@ jobs:
   test-arm-reference-delegation:
     name: test-arm-reference-delegation
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
     with:
       runner: linux.2xlarge
       docker-image: executorch-ubuntu-22.04-arm-sdk

backends/apple/coreml/test/test_coreml_partitioner.py

@@ -16,8 +16,7 @@
 
 
 class TestCoreMLPartitioner(unittest.TestCase):
-    # TODO(T182928844): Delegate dim order op to backend.
-    edge_compile_config = executorch.exir.EdgeCompileConfig(_skip_dim_order=True)
+    edge_compile_config = executorch.exir.EdgeCompileConfig()
 
     def test_add_sub_skip_mm(self):
         class Model(torch.nn.Module):

backends/apple/mps/setup.md

@@ -97,7 +97,7 @@ I 00:00:00.122615 executorch:mps_executor_runner.mm:501] Model verified successf
 ### [Optional] Run the generated model directly using pybind
 1. Make sure `pybind` MPS support was installed:
 ```bash
-./install_requirements.sh --pybind mps
+./install_executorch.sh --pybind mps
 ```
 2. Run the `mps_example` script to trace the model and run it directly from python:
 ```bash

backends/arm/_passes/annotate_channels_last_dim_order_pass.py

@@ -1,4 +1,4 @@
-# Copyright 2024 Arm Limited and/or its affiliates.
+# Copyright 2024-2025 Arm Limited and/or its affiliates.
 # All rights reserved.
 #
 # This source code is licensed under the BSD-style license found in the
@@ -15,7 +15,7 @@
     get_node_arg,
     insert_q_dq_pair,
 )
-from executorch.backends.arm.tosa_quant_utils import dq_op, q_op, register_passable_op
+from executorch.backends.arm.tosa_quant_utils import dq_op, q_op
 from executorch.backends.arm.tosa_utils import is_consumer_node_depthwise_conv2d
 from executorch.exir.dialects._ops import ops as exir_ops
 from executorch.exir.pass_base import ExportPass, PassResult
@@ -43,9 +43,6 @@ def _transpose_impl(*args, **kwargs):
     return args[0]
 
 
-register_passable_op(torch.ops.passthrough_to_tosa._transpose)
-
-
 class AnnotateChannelsLastDimOrder(ExportPass):
     """
     Annotates each node with a tosa_dim_order. tosa_dim_order can be seen as a channels-last dim-order

backends/arm/_passes/fuse_quantized_activation_pass.py

@@ -19,12 +19,13 @@ def _is_fuseable_quantized_activation(self, node: Node):
             is_fuseable = min_val == 0
 
         is_quantized = len(node.users) == 1 and next(iter(node.users)).target == q_op
-        if is_quantized:
+        if is_fuseable and is_quantized:
             quant_node = next(iter(node.users))
             zp = quant_node.args[2]
             qmin = quant_node.args[3]
-
-        return is_fuseable and is_quantized and zp == qmin
+            return zp == qmin
+        else:
+            return False
 
     def _is_fuseable_input(self, node: Node):
         return (

backends/arm/operators/__init__.py

@@ -13,7 +13,6 @@
     op_bmm,
     op_cat,
     op_conv2d,
-    op_dequant,
     op_exp,
     op_full,
     op_get_item,
@@ -24,7 +23,6 @@
     op_min,
     op_mul,
     op_permute,
-    op_quant,
     op_reciprocal,
     op_relu,
     op_repeat,