Merge branch 'main' into step_vela_to_tosa-1.0

Author: digantdesai

.Package.swift/kernels_custom/dummy.swift .Package.swift/kernels_llm/dummy.swift.Package.swift/kernels_custom/dummy.swift renamed to .Package.swift/kernels_llm/dummy.swift

@@ -1 +1 @@
-9b498d3bb28b8e3411ce464dd2755c5b96d92c8f
+7cda4017ddda554752e89069ae205be5e8388f59

…e.swift/kernels_custom_debug/dummy.swift …kage.swift/kernels_llm_debug/dummy.swift.Package.swift/kernels_custom_debug/dummy.swift renamed to .Package.swift/kernels_llm_debug/dummy.swift .Package.swift/kernels_custom_debug/dummy.swift renamed to .Package.swift/kernels_llm_debug/dummy.swift

@@ -12,4 +12,4 @@ pushd pytorch
 git checkout "$pytorch_pin"
 popd
 "$(dirname "${BASH_SOURCE[0]}")"/compare_dirs.sh runtime/core/portable_type/c10/c10 pytorch/c10
-"$(dirname "${BASH_SOURCE[0]}")"/compare_dirs.sh runtime/core/portable_type/c10/torch/standalone pytorch/torch/standalone
+"$(dirname "${BASH_SOURCE[0]}")"/compare_dirs.sh runtime/core/portable_type/c10/torch/headeronly pytorch/torch/headeronly

.ci/docker/ci_commit_pins/pytorch.txt

@@ -8,4 +8,4 @@
 # Setup arm example environment (including TOSA tools)
 git config --global user.email "github_executorch@arm.com"
 git config --global user.name "Github Executorch"
-bash examples/arm/setup.sh --i-agree-to-the-contained-eula
+bash examples/arm/setup.sh --i-agree-to-the-contained-eula ${@:-}

.ci/scripts/check_c10_sync.sh

@@ -0,0 +1,13 @@
+
+set -ex
+
+install_emscripten() {
+    git clone https://github.com/emscripten-core/emsdk.git
+    pushd emsdk || return
+    ./emsdk install 4.0.10
+    ./emsdk activate 4.0.10
+    source ./emsdk_env.sh
+    popd || return
+}
+
+install_emscripten

.ci/scripts/setup-arm-baremetal-tools.sh

@@ -35,15 +35,15 @@ cmake -DPYTHON_EXECUTABLE=python \
     -DEXECUTORCH_BUILD_XNNPACK=OFF \
     -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
-    -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+    -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
     -Bcmake-out .
 cmake --build cmake-out -j16 --target install --config Release
 
 # Install llama runner with torchao
 cmake -DPYTHON_EXECUTABLE=python \
     -DBUILD_TESTING=OFF \
     -DCMAKE_BUILD_TYPE=Release \
-    -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+    -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -DEXECUTORCH_BUILD_XNNPACK=OFF \
     -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \

.ci/scripts/setup-emscripten.sh

@@ -39,7 +39,7 @@ EXECUTORCH_COMMON_CMAKE_ARGS="                      \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
         -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
         -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON      \
-        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON        \
+        -DEXECUTORCH_BUILD_KERNELS_LLM=ON        \
         -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON     \
         -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON     \
         -DEXECUTORCH_BUILD_XNNPACK=ON               \
@@ -69,7 +69,7 @@ LLAVA_COMMON_CMAKE_ARGS="                        \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -DCMAKE_INSTALL_PREFIX=${BUILD_DIR}      \
         -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}   \
-        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON     \
+        -DEXECUTORCH_BUILD_KERNELS_LLM=ON     \
         -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON  \
         -DEXECUTORCH_BUILD_XNNPACK=ON"
 

.ci/scripts/test_llama_torchao_lowbit.sh

@@ -188,22 +188,6 @@ test_model_with_qnn() {
     EXPORT_SCRIPT=edsr
     # Additional deps for edsr
     pip install piq
-  elif [[ "${MODEL_NAME}" == "cvt" ]]; then
-    EXPORT_SCRIPT=cvt
-  elif [[ "${MODEL_NAME}" == "dit" ]]; then
-    EXPORT_SCRIPT=dit
-  elif [[ "${MODEL_NAME}" == "efficientnet" ]]; then
-    EXPORT_SCRIPT=efficientnet
-  elif [[ "${MODEL_NAME}" == "focalnet" ]]; then
-    EXPORT_SCRIPT=focalnet
-  elif [[ "${MODEL_NAME}" == "mobilevit_v1" ]]; then
-    EXPORT_SCRIPT=mobilevit_v1
-  elif [[ "${MODEL_NAME}" == "mobilevit_v2" ]]; then
-    EXPORT_SCRIPT=mobilevit_v2
-  elif [[ "${MODEL_NAME}" == "pvt" ]]; then
-    EXPORT_SCRIPT=pvt
-  elif [[ "${MODEL_NAME}" == "swin" ]]; then
-    EXPORT_SCRIPT=swin_transformer
   elif [[ "${MODEL_NAME}" == "albert" ]]; then
     EXPORT_SCRIPT=albert
   elif [[ "${MODEL_NAME}" == "bert" ]]; then
@@ -212,8 +196,6 @@ test_model_with_qnn() {
     EXPORT_SCRIPT=distilbert
   elif [[ "${MODEL_NAME}" == "eurobert" ]]; then
     EXPORT_SCRIPT=eurobert
-  elif [[ "${MODEL_NAME}" == "roberta" ]]; then
-    EXPORT_SCRIPT=roberta
   else
     echo "Unsupported model $MODEL_NAME"
     exit 1
@@ -228,13 +210,10 @@ test_model_with_qnn() {
     "dl3"|"mv3"|"mv2"|"ic4"|"ic3"|"vit"|"mb"|"w2l")
         SCRIPT_FOLDER=scripts
         ;;
-    "cvt"|"dit"|"focalnet"|"mobilevit_v2"|"pvt"|"swin")
-        SCRIPT_FOLDER=oss_scripts
-        ;;
-    "albert"|"bert"|"distilbert"|"roberta"|"efficientnet"|"mobilevit_v1")
+    "albert"|"bert"|"distilbert")
         pip install evaluate
         SCRIPT_FOLDER=oss_scripts
-        # 16bit models will encounter op validation fail on some operations,
+        # Bert models running in 16bit will encounter op validation fail on some operations,
         # which requires CHIPSET >= SM8550.
         QNN_CHIPSET=SM8550
         ;;

.ci/scripts/test_llava.sh

@@ -33,7 +33,7 @@ cmake_install_executorch_libraries() {
       -DEXECUTORCH_BUILD_XNNPACK=ON \
       -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
       -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
-      -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+      -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
       -B${BUILD_DIR} .
 
   cmake --build ${BUILD_DIR} -j${NPROC} --target install --config ${BUILD_TYPE}
@@ -43,7 +43,7 @@ cmake_build_phi_3_mini() {
   cmake -DPYTHON_EXECUTABLE=$PYTHON_EXECUTABLE \
       -DCMAKE_INSTALL_PREFIX=${BUILD_DIR} \
       -DCMAKE_BUILD_TYPE=${BUILD_TYPE} \
-      -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+      -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
       -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
       -DEXECUTORCH_BUILD_XNNPACK=ON \
       -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \