Merge branch 'main' into add-dim-order-clone-aot

Author: Gasoonjia

.ci/docker/ci_commit_pins/optimum-executorch.txt

@@ -1 +1 @@
-a3942627f5ac048e06b4b1d703b0a6a53bf6da5b
+eea657ddbdeb1118943a92fb73c289985c3ee1ba

.ci/docker/ci_commit_pins/pytorch.txt

@@ -1 +1 @@
-ab43fe4bdf5ccd82897f0e982c451a0127bd175e
+6fc0ad22f0a07b6f38d138861c56a765d5a9bb02

.ci/docker/common/install_zephyr.sh

@@ -8,6 +8,9 @@
 
 set -ex
 
+# shellcheck source=/dev/null
+source "$(dirname "${BASH_SOURCE[0]}")/utils.sh"
+
 # Double check if the NDK version is set
 [ -n "${ZEPHYR_SDK}" ]
 
@@ -77,16 +80,8 @@ install_prerequiresites() {
         chmod +x kitware-archive.sh && \
         ./kitware-archive.sh && \
         rm -f kitware-archive.sh
-    useradd -d /home/zephyruser -m -s /bin/bash zephyruser
-}
-
-install_sdk() {
-    wget https://github.com/zephyrproject-rtos/sdk-ng/releases/download/v0.16.0/zephyr-sdk-0.16.0_linux-x86_64.tar.xz && \
-         tar -xf zephyr-sdk-0.16.0_linux-x86_64.tar.xz && \
-         rm -f zephyr-sdk-0.16.0_linux-x86_64.tar.xz && \
-         cd zephyr-sdk-0.16.0/ && \
-         ./setup.sh -c -t arm-zephyr-eabi
+    pip_install --no-cache-dir west
+    pip_install pyelftools
 }
 
 install_prerequiresites
-install_sdk

.ci/docker/ubuntu/Dockerfile

@@ -86,8 +86,9 @@ ARG ARM_SDK
 
 ARG ZEPHYR_SDK
 COPY ./common/install_zephyr.sh install_zephyr.sh
+COPY ./common/utils.sh utils.sh
 RUN if [ -n "${ZEPHYR_SDK}" ]; then bash ./install_zephyr.sh; fi
-RUN rm install_zephyr.sh
+RUN rm install_zephyr.sh utils.sh
 
 ARG QNN_SDK
 

.ci/scripts/build-qnn-sdk.sh

@@ -33,6 +33,8 @@ set_up_aot() {
       -DEXECUTORCH_BUILD_DEVTOOLS=ON \
       -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
       -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
+      -DEXECUTORCH_BUILD_EXTENSION_EXTENSION_LLM=ON \
+      -DEXECUTORCH_BUILD_EXTENSION_EXTENSION_LLM_RUNNER=ON \
       -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
       -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON \
       -DEXECUTORCH_ENABLE_EVENT_TRACER=ON \

.ci/scripts/setup-emscripten.sh

@@ -7,6 +7,13 @@
 
 set -ex
 
+# need version >= 17
+install_node() {
+    curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash
+    source "$HOME/.nvm/nvm.sh"
+    nvm install 22
+}
+
 install_emscripten() {
     git clone https://github.com/emscripten-core/emsdk.git
     pushd emsdk || return
@@ -16,4 +23,5 @@ install_emscripten() {
     popd || return
 }
 
+install_node
 install_emscripten

.ci/scripts/setup-qnn-deps.sh

@@ -7,47 +7,7 @@
 
 set -ex
 
-verify_pkg_installed() {
-  echo $(dpkg-query -W --showformat='${Status}\n' $1|grep "install ok installed")
-}
+source "$(dirname "${BASH_SOURCE[0]}")/../../backends/qualcomm/scripts/install_qnn_sdk.sh"
 
-install_qnn() {
-  echo "Start installing qnn."
-  QNN_INSTALLATION_DIR=/tmp/qnn
-  mkdir -p "${QNN_INSTALLATION_DIR}"
-
-  curl -Lo /tmp/v2.28.0.24.10.29.zip "https://softwarecenter.qualcomm.com/api/download/software/qualcomm_neural_processing_sdk/v2.28.0.241029.zip"
-  echo "Finishing downloading qnn sdk."
-  unzip -qo /tmp/v2.28.0.24.10.29.zip -d /tmp
-  echo "Finishing unzip qnn sdk."
-
-
-  # Print the content for manual verification
-  ls -lah "/tmp/qairt"
-  mv "/tmp/qairt"/* "${QNN_INSTALLATION_DIR}"
-  echo "Finishing installing qnn '${QNN_INSTALLATION_DIR}' ."
-
-  ls -lah "${QNN_INSTALLATION_DIR}"
-}
-
-setup_libc++() {
-  clang_version=$1
-  sudo apt-get update
-  pkgs_to_check=("libc++-${clang_version}-dev")
-  j=0
-  while [ $j -lt ${#pkgs_to_check[*]} ]; do
-    install_status=$(verify_pkg_installed ${pkgs_to_check[$j]})
-    if [ "$install_status" == "" ]; then
-      sudo apt-get install -y ${pkgs_to_check[$j]}
-      if [[ $? -ne 0 ]]; then
-        echo "ERROR: Failed to install required packages for libc++"
-        exit 1
-      fi
-    fi
-    j=$(( $j +1));
-  done
-}
-
-# This needs to match with the clang version from the Docker image
-setup_libc++ 12
+setup_libcpp 12
 install_qnn

.ci/scripts/test_ane_static_llama.sh

@@ -28,6 +28,6 @@ pushd $EXECUTORCH_ROOT/examples/apple/coreml/llama
 # Download stories llama110m artifacts
 download_stories_model_artifacts
 
-python export.py -n model.pte -p params.json -c stories110M.pt --seq_length 32 --max_seq_length 64 --dtype fp16 --coreml-quantize c4w
+python export.py -n model.pte -p params.json -c stories110M.pt --seq_length 32 --max_seq_length 64 --dtype fp16 --coreml-quantize c4w --embedding-quantize 4,32
 
 popd

.ci/scripts/test_llama.sh

@@ -150,6 +150,7 @@ cmake_install_executorch_libraries() {
     echo "Installing libexecutorch.a, libextension_module.so, libportable_ops_lib.a"
     rm -rf cmake-out
     retry cmake --preset llm \
+        -DBUILD_TESTING=OFF \
         -DCMAKE_INSTALL_PREFIX=cmake-out \
         -DCMAKE_BUILD_TYPE="$CMAKE_BUILD_TYPE" \
         -DEXECUTORCH_BUILD_QNN="$QNN" \

.ci/scripts/test_llama_torchao_lowbit.sh

@@ -29,6 +29,8 @@ cmake -DPYTHON_EXECUTABLE=python \
     -DEXECUTORCH_ENABLE_LOGGING=1 \
     -DCMAKE_BUILD_TYPE=Release \
     -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
+    -DEXECUTORCH_BUILD_EXTENSION_LLM=ON \
+    -DEXECUTORCH_BUILD_EXTENSION_LLM_RUNNER=ON \
     -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
     -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
     -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON \