Add suffixes to cmake flags related to building kernels. (pytorch#3499)

Summary:
Pull Request resolved: pytorch#3499

.

Reviewed By: cccclai

Differential Revision: D56917214

fbshipit-source-id: 6dbb4c1f5c981e7da9215469d6e38a90bc731515

.ci/scripts/build_llama_android.sh

@@ -25,8 +25,8 @@ install_executorch_and_backend_lib() {
     -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
     -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
     -DEXECUTORCH_BUILD_XNNPACK=ON \
-    -DEXECUTORCH_BUILD_OPTIMIZED=ON \
-    -DEXECUTORCH_BUILD_QUANTIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
     -DXNNPACK_ENABLE_ARM_BF16=OFF \
     -Bcmake-android-out .
 
@@ -42,7 +42,7 @@ build_llama_runner() {
     -DANDROID_PLATFORM=android-23 \
     -DCMAKE_INSTALL_PREFIX=cmake-android-out \
     -DCMAKE_BUILD_TYPE=Release -DPYTHON_EXECUTABLE=python \
-    -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -Bcmake-android-out/examples/models/llama2 examples/models/llama2
 
     cmake --build cmake-android-out/examples/models/llama2 -j4 --config Release

.ci/scripts/test_llama.sh

@@ -73,9 +73,9 @@ cmake_install_executorch_libraries() {
         -DCMAKE_BUILD_TYPE=Debug \
         -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
-        -DEXECUTORCH_BUILD_CUSTOM="$CUSTOM" \
-        -DEXECUTORCH_BUILD_OPTIMIZED=ON \
-        -DEXECUTORCH_BUILD_QUANTIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_CUSTOM="$CUSTOM" \
+        -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
         -DEXECUTORCH_BUILD_XNNPACK="$XNNPACK" \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -Bcmake-out .
@@ -88,8 +88,8 @@ cmake_build_llama_runner() {
     retry cmake -DBUCK2="$BUCK" \
         -DCMAKE_INSTALL_PREFIX=cmake-out \
         -DCMAKE_BUILD_TYPE=Debug \
-        -DEXECUTORCH_BUILD_CUSTOM="$CUSTOM" \
-        -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_CUSTOM="$CUSTOM" \
+        -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
         -DEXECUTORCH_BUILD_XNNPACK="$XNNPACK" \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -Bcmake-out/${dir} \

.ci/scripts/test_quantized_aot_lib.sh

@@ -23,7 +23,7 @@ build_cmake_quantized_aot_lib() {
     && cd ${CMAKE_OUTPUT_DIR} \
     && retry cmake -DCMAKE_BUILD_TYPE=Release \
       -DCMAKE_PREFIX_PATH="$CMAKE_PREFIX_PATH" \
-      -DEXECUTORCH_BUILD_QUANTIZED_OPS_AOT=ON \
+      -DEXECUTORCH_BUILD_KERNELS_QUANTIZED_AOT=ON \
       -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" ..)
 
   cmake --build ${CMAKE_OUTPUT_DIR} -j4

CMakeLists.txt

@@ -149,9 +149,11 @@ option(EXECUTORCH_BUILD_ARM_BAREMETAL
 
 option(EXECUTORCH_BUILD_COREML "Build the Core ML backend" OFF)
 
-option(EXECUTORCH_BUILD_CUSTOM "Build the custom kernels" OFF)
+option(EXECUTORCH_BUILD_KERNELS_CUSTOM "Build the custom kernels" OFF)
 
-option(EXECUTORCH_BUILD_CUSTOM_OPS_AOT "Build the custom ops lib for AOT" OFF)
+option(EXECUTORCH_BUILD_KERNELS_CUSTOM_AOT "Build the custom ops lib for AOT"
+       OFF
+)
 
 option(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER "Build the Data Loader extension"
        OFF
@@ -171,9 +173,9 @@ option(EXECUTORCH_BUILD_PYBIND "Build the Python Bindings" OFF)
 
 option(EXECUTORCH_BUILD_QNN "Build the Qualcomm backend" OFF)
 
-option(EXECUTORCH_BUILD_OPTIMIZED "Build the optimized kernels" OFF)
+option(EXECUTORCH_BUILD_KERNELS_OPTIMIZED "Build the optimized kernels" OFF)
 
-option(EXECUTORCH_BUILD_QUANTIZED "Build the quantized kernels" OFF)
+option(EXECUTORCH_BUILD_KERNELS_QUANTIZED "Build the quantized kernels" OFF)
 
 option(EXECUTORCH_BUILD_SDK "Build the ExecuTorch SDK")
 
@@ -199,12 +201,12 @@ cmake_dependent_option(
   "NOT EXECUTORCH_BUILD_ARM_BAREMETAL" OFF
 )
 
-if(EXECUTORCH_BUILD_CUSTOM_OPS_AOT)
-  set(EXECUTORCH_BUILD_CUSTOM ON)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM_AOT)
+  set(EXECUTORCH_BUILD_KERNELS_CUSTOM ON)
 endif()
 
-if(EXECUTORCH_BUILD_CUSTOM)
-  set(EXECUTORCH_BUILD_OPTIMIZED ON)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM)
+  set(EXECUTORCH_BUILD_KERNELS_OPTIMIZED ON)
 endif()
 
 if(EXECUTORCH_BUILD_CPUINFO)
@@ -443,18 +445,18 @@ target_link_options_shared_lib(executorch)
 #
 add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/kernels/portable)
 
-if(EXECUTORCH_BUILD_CUSTOM)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM)
   # TODO: move all custom kernels to ${CMAKE_CURRENT_SOURCE_DIR}/kernels/custom
   add_subdirectory(
     ${CMAKE_CURRENT_SOURCE_DIR}/examples/models/llama2/custom_ops
   )
 endif()
 
-if(EXECUTORCH_BUILD_OPTIMIZED)
+if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
   add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/kernels/optimized)
 endif()
 
-if(EXECUTORCH_BUILD_QUANTIZED)
+if(EXECUTORCH_BUILD_KERNELS_QUANTIZED)
   add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/kernels/quantized)
 endif()
 
@@ -492,14 +494,14 @@ if(EXECUTORCH_BUILD_EXECUTOR_RUNNER)
   # Baseline libraries that executor_runner will link against.
   set(_executor_runner_libs executorch gflags)
 
-  if(EXECUTORCH_BUILD_OPTIMIZED)
+  if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
     list(APPEND _executor_runner_libs optimized_native_cpu_ops_lib)
   else()
     list(APPEND _executor_runner_libs portable_ops_lib)
   endif()
 
   # Generate lib to register quantized ops
-  if(EXECUTORCH_BUILD_QUANTIZED)
+  if(EXECUTORCH_BUILD_KERNELS_QUANTIZED)
     list(APPEND _executor_runner_libs quantized_ops_lib)
   endif()
 
@@ -606,19 +608,19 @@ if(EXECUTORCH_BUILD_PYBIND)
     list(APPEND _dep_libs xnnpack_backend XNNPACK)
   endif()
 
-  if(EXECUTORCH_BUILD_QUANTIZED)
+  if(EXECUTORCH_BUILD_KERNELS_QUANTIZED)
     target_link_options_shared_lib(quantized_ops_lib)
     list(APPEND _dep_libs quantized_kernels quantized_ops_lib)
   endif()
 
   # TODO(larryliu): Fix macOS 2 dylibs having 2 sets of static variables issue
-  if(EXECUTORCH_BUILD_CUSTOM_OPS_AOT AND NOT APPLE)
+  if(EXECUTORCH_BUILD_KERNELS_CUSTOM_AOT AND NOT APPLE)
     list(APPEND _dep_libs custom_ops_aot_lib)
   endif()
   # TODO(laryliu): Fix linux duplicate registation problem. In GH CI worker
   # libcustom_ops.a doesn't dedup with the one indirectly linked from
   # libcustom_ops_aot_lib.a
-  if(EXECUTORCH_BUILD_CUSTOM AND APPLE)
+  if(EXECUTORCH_BUILD_KERNELS_CUSTOM AND APPLE)
     target_link_options_shared_lib(custom_ops)
     list(APPEND _dep_libs custom_ops)
   endif()

backends/vulkan/docs/android_demo.md

@@ -97,7 +97,7 @@ binary using the Android NDK toolchain.
     -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
     -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
     -DEXECUTORCH_BUILD_VULKAN=ON \
-    -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -DPYTHON_EXECUTABLE=python \
     -Bcmake-android-out && \
   cmake --build cmake-android-out -j16 --target install)

build/Utils.cmake

@@ -50,8 +50,8 @@ function(executorch_print_configuration_summary)
     STATUS
       "  EXECUTORCH_BUILD_COREML                : ${EXECUTORCH_BUILD_COREML}"
   )
-  message(STATUS "  EXECUTORCH_BUILD_CUSTOM                : "
-                 "${EXECUTORCH_BUILD_CUSTOM}"
+  message(STATUS "  EXECUTORCH_BUILD_KERNELS_CUSTOM        : "
+                 "${EXECUTORCH_BUILD_KERNELS_CUSTOM}"
   )
   message(STATUS "  EXECUTORCH_BUILD_EXECUTOR_RUNNER       : "
                  "${EXECUTORCH_BUILD_EXECUTOR_RUNNER}"
@@ -90,11 +90,11 @@ function(executorch_print_configuration_summary)
   message(
     STATUS "  EXECUTORCH_BUILD_QNN                   : ${EXECUTORCH_BUILD_QNN}"
   )
-  message(STATUS "  EXECUTORCH_BUILD_OPTIMIZED             : "
-                 "${EXECUTORCH_BUILD_OPTIMIZED}"
+  message(STATUS "  EXECUTORCH_BUILD_KERNELS_OPTIMIZED     : "
+                 "${EXECUTORCH_BUILD_KERNELS_OPTIMIZED}"
   )
-  message(STATUS "  EXECUTORCH_BUILD_QUANTIZED             : "
-                 "${EXECUTORCH_BUILD_QUANTIZED}"
+  message(STATUS "  EXECUTORCH_BUILD_KERNELS_QUANTIZED     : "
+                 "${EXECUTORCH_BUILD_KERNELS_QUANTIZED}"
   )
   message(
     STATUS "  EXECUTORCH_BUILD_SDK                   : ${EXECUTORCH_BUILD_SDK}"

build/build_apple_frameworks.sh

@@ -165,9 +165,9 @@ cmake_build() {
         -DEXECUTORCH_BUILD_EXTENSION_APPLE=ON \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
         -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-        -DEXECUTORCH_BUILD_CUSTOM=$CUSTOM \
-        -DEXECUTORCH_BUILD_OPTIMIZED=$OPTIMIZED \
-        -DEXECUTORCH_BUILD_QUANTIZED=$QUANTIZED \
+        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=$CUSTOM \
+        -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=$OPTIMIZED \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=$QUANTIZED \
         -DCMAKE_ARCHIVE_OUTPUT_DIRECTORY="$(pwd)" \
         ${platform_flag:+-DPLATFORM=$platform_flag} \
         ${platform_target:+-DDEPLOYMENT_TARGET=$platform_target} \

configurations/CMakeLists.txt

@@ -31,7 +31,7 @@ set(_common_compile_options -Wno-deprecated-declarations)
 include(${EXECUTORCH_ROOT}/build/Utils.cmake)
 include(${EXECUTORCH_ROOT}/build/Codegen.cmake)
 
-if(EXECUTORCH_BUILD_OPTIMIZED)
+if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
   # Merge optimized and portable definitions, taking optimized where available.
   merge_yaml(
     FUNCTIONS_YAML ${EXECUTORCH_ROOT}/kernels/optimized/optimized-oss.yaml

docs/source/llm/getting-started.md

@@ -367,7 +367,7 @@ set(CMAKE_CXX_STANDARD_REQUIRED True)
 # Set options for executorch build.
 option(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER "" ON)
 option(EXECUTORCH_BUILD_EXTENSION_MODULE "" ON)
-option(EXECUTORCH_BUILD_OPTIMIZED "" ON)
+option(EXECUTORCH_BUILD_KERNELS_OPTIMIZED "" ON)
 
 # Include the executorch subdirectory.
 add_subdirectory(
@@ -521,7 +521,7 @@ set(CMAKE_CXX_STANDARD_REQUIRED True)
 # Set options for executorch build.
 option(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER "" ON)
 option(EXECUTORCH_BUILD_EXTENSION_MODULE "" ON)
-option(EXECUTORCH_BUILD_OPTIMIZED "" ON)
+option(EXECUTORCH_BUILD_KERNELS_OPTIMIZED "" ON)
 option(EXECUTORCH_BUILD_XNNPACK "" ON) # Build with Xnnpack backend
 
 # Include the executorch subdirectory.

examples/arm/run.sh

@@ -73,8 +73,8 @@ function build_quantization_aot_lib()
     cmake -DBUCK2=${buck2} \
         -DCMAKE_BUILD_TYPE=Release \
         -DEXECUTORCH_BUILD_XNNPACK=OFF \
-        -DEXECUTORCH_BUILD_QUANTIZED=ON \
-        -DEXECUTORCH_BUILD_QUANTIZED_OPS_AOT=ON \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED_AOT=ON \
         -DCMAKE_PREFIX_PATH="$CMAKE_PREFIX_PATH" \
         -DPYTHON_EXECUTABLE=python3 \
 	-Bcmake-out-aot-lib \
@@ -101,7 +101,7 @@ function build_executorch() {
         -DCMAKE_BUILD_TYPE=Release                        \
         -DEXECUTORCH_ENABLE_LOGGING=ON                    \
         -DEXECUTORCH_BUILD_ARM_BAREMETAL=ON               \
-        -DEXECUTORCH_BUILD_QUANTIZED=ON                   \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON                   \
         -DEXECUTORCH_BUILD_EXTENSION_RUNNER_UTIL=ON       \
         -DFLATC_EXECUTABLE="$(which flatc)"               \
         -DCMAKE_TOOLCHAIN_FILE="${toolchain_cmake}"       \

examples/demo-apps/android/ExecuTorchDemo/setup.sh

@@ -15,7 +15,7 @@ cmake . -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
   -DEXECUTORCH_BUILD_XNNPACK=ON \
   -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
   -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-  -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+  -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
   -DCMAKE_BUILD_TYPE=Release \
   -B"${CMAKE_OUT}"
 

examples/demo-apps/android/LlamaDemo/setup-with-qnn.sh

@@ -15,7 +15,7 @@ cmake . -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
   -DEXECUTORCH_BUILD_XNNPACK=ON \
   -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
   -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-  -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+  -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
   -DEXECUTORCH_BUILD_QNN=ON \
   -DQNN_SDK_ROOT="${QNN_SDK_ROOT}" \
   -DCMAKE_BUILD_TYPE=Release \

examples/demo-apps/android/LlamaDemo/setup.sh

@@ -16,9 +16,9 @@ cmake . -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
   -DEXECUTORCH_BUILD_XNNPACK=ON \
   -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
   -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-  -DEXECUTORCH_BUILD_OPTIMIZED=ON \
-  -DEXECUTORCH_BUILD_QUANTIZED=ON \
-  -DEXECUTORCH_BUILD_CUSTOM=ON \
+  -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
+  -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
+  -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
   -DCMAKE_BUILD_TYPE=Release \
   -B"${CMAKE_OUT}"
 
@@ -34,8 +34,8 @@ cmake examples/models/llama2 \
          -DANDROID_ABI="$ANDROID_ABI" \
          -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
          -DEXECUTORCH_USE_TIKTOKEN="${EXECUTORCH_USE_TIKTOKEN}" \
-         -DEXECUTORCH_BUILD_CUSTOM=ON \
-         -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+         -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+         -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
          -DEXECUTORCH_BUILD_XNNPACK=ON \
          -DCMAKE_BUILD_TYPE=Release \
          -B"${CMAKE_OUT}"/examples/models/llama2

examples/llm_manual/CMakeLists.txt

@@ -13,7 +13,7 @@ set(CMAKE_CXX_STANDARD_REQUIRED True)
 # Set options for executorch build.
 option(EXECUTORCH_BUILD_EXTENSION_DATA_LOADER "" ON)
 option(EXECUTORCH_BUILD_EXTENSION_MODULE "" ON)
-option(EXECUTORCH_BUILD_OPTIMIZED "" ON)
+option(EXECUTORCH_BUILD_KERNELS_OPTIMIZED "" ON)
 option(EXECUTORCH_BUILD_XNNPACK "" ON) # Build with Xnnpack backend
 
 # Include the executorch subdirectory.

examples/models/llama2/CMakeLists.txt

@@ -19,7 +19,7 @@ cmake_minimum_required(VERSION 3.19)
 project(llama_runner)
 
 # Duplicating options as root CMakeLists.txt
-option(EXECUTORCH_BUILD_OPTIMIZED "Build the optimized kernels" OFF)
+option(EXECUTORCH_BUILD_KERNELS_OPTIMIZED "Build the optimized kernels" OFF)
 
 option(EXECUTORCH_USE_TIKTOKEN "Use Tiktoken as a tokenizer" OFF)
 
@@ -85,7 +85,7 @@ if(CMAKE_TOOLCHAIN_IOS OR ANDROID)
 endif()
 
 # custom ops library
-if(EXECUTORCH_BUILD_CUSTOM)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM)
   add_subdirectory(custom_ops)
 endif()
 
@@ -106,7 +106,7 @@ endif()
 set(link_libraries gflags)
 set(_srcs main.cpp)
 
-if(EXECUTORCH_BUILD_OPTIMIZED)
+if(EXECUTORCH_BUILD_KERNELS_OPTIMIZED)
   list(
     APPEND
     link_libraries
@@ -126,7 +126,7 @@ endif()
 target_link_options_shared_lib(quantized_ops_lib)
 list(APPEND link_libraries quantized_kernels quantized_ops_lib)
 
-if(EXECUTORCH_BUILD_CUSTOM)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM)
   target_link_options_shared_lib(custom_ops)
   list(APPEND link_libraries custom_ops)
 endif()

examples/models/llama2/README.md

@@ -161,9 +161,9 @@ The Wikitext results generated above used: `{max_seq_len: 2048, limit: 1000}`
         -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
         -DEXECUTORCH_BUILD_XNNPACK=ON \
-        -DEXECUTORCH_BUILD_QUANTIZED=ON \
-        -DEXECUTORCH_BUILD_OPTIMIZED=ON \
-        -DEXECUTORCH_BUILD_CUSTOM=ON \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
         -Bcmake-out .
 
     cmake --build cmake-out -j16 --target install --config Release
@@ -174,10 +174,10 @@ The Wikitext results generated above used: `{max_seq_len: 2048, limit: 1000}`
     cmake -DPYTHON_EXECUTABLE=python \
         -DCMAKE_INSTALL_PREFIX=cmake-out \
         -DCMAKE_BUILD_TYPE=Release \
-        -DEXECUTORCH_BUILD_CUSTOM=ON \
-        -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+        -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
         -DEXECUTORCH_BUILD_XNNPACK=ON \
-        -DEXECUTORCH_BUILD_QUANTIZED=ON \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
         -Bcmake-out/examples/models/llama2 \
         examples/models/llama2
 
@@ -215,7 +215,7 @@ cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake \
     -DEXECUTORCH_ENABLE_LOGGING=1 \
     -DEXECUTORCH_BUILD_XNNPACK=ON \
     -DPYTHON_EXECUTABLE=python \
-    -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -Bcmake-out-android .
 
 cmake --build cmake-out-android -j16 --target install --config Release
@@ -229,7 +229,7 @@ cmake  -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake \
     -DCMAKE_INSTALL_PREFIX=cmake-out-android \
     -DCMAKE_BUILD_TYPE=Release \
     -DPYTHON_EXECUTABLE=python \
-    -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+    -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -Bcmake-out-android/examples/models/llama2 \
     examples/models/llama2
 

examples/models/llama2/custom_ops/CMakeLists.txt

@@ -80,7 +80,7 @@ target_compile_options(
 
 install(TARGETS custom_ops DESTINATION lib)
 
-if(EXECUTORCH_BUILD_CUSTOM_OPS_AOT)
+if(EXECUTORCH_BUILD_KERNELS_CUSTOM_AOT)
   # Add a AOT library
   find_package(Torch CONFIG REQUIRED)
   add_library(