Android demo app tutorial fix for XNNPACK and QNN (#2962) (#3027)

Summary:
* Update tutorial due to recent changes.
* Clean up setup.sh for app helper lib build.

Pull Request resolved: #2962

Reviewed By: cccclai

Differential Revision: D55951189

Pulled By: kirklandsign

fbshipit-source-id: 2c95e8580145b039f503e7cd99a4003867f8dbb0
(cherry picked from commit 26365f1)

Author: kirklandsign

examples/demo-apps/android/ExecuTorchDemo/README.md

@@ -17,7 +17,7 @@ This guide explains how to setup ExecuTorch for Android using a demo app. The ap
 * Refer to [Setting up ExecuTorch](https://pytorch.org/executorch/stable/getting-started-setup) to set up the repo and dev environment.
 * Download and install [Android Studio and SDK](https://developer.android.com/studio).
 * Supported Host OS: CentOS, macOS Ventura (M1/x86_64). See below for Qualcomm HTP specific requirements.
-* *Qualcomm HTP Only[^1]:* To build and run on Qualcomm's AI Engine Direct, please follow [Building and Running ExecuTorch with Qualcomm AI Engine Direct Backend](build-run-qualcomm-ai-engine-direct-backend.md) for hardware and software pre-requisites.
+* *Qualcomm HTP Only[^1]:* To build and run on Qualcomm's AI Engine Direct, please follow [Building and Running ExecuTorch with Qualcomm AI Engine Direct Backend](build-run-qualcomm-ai-engine-direct-backend.md) for hardware and software pre-requisites. The version we use for this tutorial is 2.19. The chip we use for this tutorial is SM8450.
 :::
 ::::
 
@@ -39,7 +39,7 @@ We generate the model file for the ExecuTorch runtime in Android Demo App.
 For delegating DeepLab v3 to XNNPACK backend, please do the following to export the model:
 
 ```bash
-export FLATC_EXECUTABLE=$(realpath third-party/flatbuffers/cmake-out/flatc)
+export FLATC_EXECUTABLE=$(realpath third-party/flatbuffers/cmake-android-out/flatc)
 python3 -m examples.xnnpack.aot_compiler --model_name="dl3" --delegate
 mkdir -p examples/demo-apps/android/ExecuTorchDemo/app/src/main/assets/
 cp dl3_xnnpack_fp32.pte examples/demo-apps/android/ExecuTorchDemo/app/src/main/assets/
@@ -54,7 +54,7 @@ For delegating to Qualcomm Hexagon NPU, please follow the tutorial [here](build-
 After generating the model, copy the model to `assets` directory.
 
 ```bash
-python -m examples.qualcomm.scripts.deeplab_v3 -b build_android -m SM8550 -s <adb_connected_device_serial>
+python -m examples.qualcomm.scripts.deeplab_v3 -b build_android -m SM8450 -s <adb_connected_device_serial>
 cp deeplab_v3/dlv3_qnn.pte examples/demo-apps/android/ExecuTorchDemo/app/src/main/assets/
 ```
 
@@ -68,22 +68,20 @@ We build the required ExecuTorch runtime library to run the model.
 
 ```bash
 export ANDROID_NDK=<path-to-android-ndk>
-export BUCK2=/tmp/buck2 # Or your buck path
+export ANDROID_ABI=arm64-v8a
 
-rm -rf cmake-out && mkdir cmake-out && cd cmake-out
+rm -rf cmake-android-out && mkdir cmake-android-out
 
 # Build the core executorch library
-cmake .. -DCMAKE_INSTALL_PREFIX=cmake-out \
+cmake . -DCMAKE_INSTALL_PREFIX=cmake-android-out \
   -DCMAKE_TOOLCHAIN_FILE="${ANDROID_NDK}/build/cmake/android.toolchain.cmake" \
   -DANDROID_ABI="${ANDROID_ABI}" \
-  -DBUCK2="${BUCK2}" \
   -DEXECUTORCH_BUILD_XNNPACK=ON \
-  -DEXECUTORCH_BUILD_FLATC=OFF \
   -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
-  -DFLATC_EXECUTABLE="${FLATC}" \
-  -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON
+  -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
+  -Bcmake-android-out
 
-cmake --build . -j16 --target install
+cmake --build cmake-android-out -j16 --target install
 ```
 
 When we set `EXECUTORCH_BUILD_XNNPACK=ON`, we will build the target [`xnnpack_backend`](https://github.com/pytorch/executorch/blob/main/backends/xnnpack/CMakeLists.txt) which in turn is linked into libexecutorch_jni via [CMake](https://github.com/pytorch/executorch/blob/main/examples/demo-apps/android/jni/CMakeLists.txt).
@@ -93,60 +91,63 @@ When we set `EXECUTORCH_BUILD_XNNPACK=ON`, we will build the target [`xnnpack_ba
 ```bash
 
 # Build the android extension
-cmake ../extension/android -DBUCK2="${BUCK2}" \
-  -DCMAKE_TOOLCHAIN_FILE=${ANDROID_NDK}/build/cmake/android.toolchain.cmake \
+cmake extension/android \
+  -DCMAKE_TOOLCHAIN_FILE="${ANDROID_NDK}"/build/cmake/android.toolchain.cmake \
   -DANDROID_ABI="${ANDROID_ABI}" \
-  -DCMAKE_INSTALL_PREFIX=cmake-out \
-  -Bextension/android
+  -DCMAKE_INSTALL_PREFIX=cmake-android-out \
+  -Bcmake-android-out/extension/android
 
-cmake --build ./extension/android -j16
+cmake --build cmake-android-out/extension/android -j16
 ```
 
 `libexecutorch_jni.so` wraps up the required XNNPACK Backend runtime library from `xnnpack_backend`, and adds an additional JNI layer using fbjni. This is later exposed to Java app.
 
 #### Qualcomm Hexagon NPU
 
-1. Configure the CMake target for the library with Qualcomm Hexagon NPU (HTP) backend (XNNPACK also included):
+1. Build the CMake target for the library with Qualcomm Hexagon NPU (HTP) backend (XNNPACK also included):
 
 ```bash
 export ANDROID_NDK=<path-to-android-ndk>
-export QNN_SDK=<path-to-qnn-sdk>
-
-rm -rf cmake-out && mkdir cmake-out && cd cmake-out
-cmake .. \
-    -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake \
-    -DANDROID_ABI=arm64-v8a \
-    -DBUCK2=/tmp/buck2 \
-    -DEXECUTORCH_BUILD_ANDROID_JNI=ON \
+export ANDROID_ABI=arm64-v8a
+export QNN_SDK_ROOT=<path-to-qnn-sdk>
+
+rm -rf cmake-android-out && mkdir cmake-android-out && cd cmake-android-out
+cmake . -DCMAKE_INSTALL_PREFIX=cmake-android-out \
+    -DCMAKE_TOOLCHAIN_FILE="${ANDROID_NDK}/build/cmake/android.toolchain.cmake" \
+    -DANDROID_ABI="${ANDROID_ABI}" \
     -DEXECUTORCH_BUILD_XNNPACK=ON \
-    -DEXECUTORCH_BUILD_FLATC=OFF \
     -DEXECUTORCH_BUILD_QNN=ON \
-    -DQNN_SDK_ROOT=$QNN_SDK \
-    -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON
+    -DQNN_SDK_ROOT="${QNN_SDK_ROOT}" \
+    -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
+    -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
+    -Bcmake-android-out
+
+cmake --build cmake-android-out -j16 --target install
 ```
 Similar to the XNNPACK library, with this setup, we compile `libexecutorch_jni.so` but it adds an additional static library `qnn_executorch_backend` which wraps up Qualcomm HTP runtime library and registers the Qualcomm HTP backend. This is later exposed to Java app.
 
 `qnn_executorch_backend` is built when we turn on CMake option `EXECUTORCH_BUILD_QNN`. It will include the [CMakeLists.txt](https://github.com/pytorch/executorch/blob/main/backends/qualcomm/CMakeLists.txt) from backends/qualcomm where we `add_library(qnn_executorch_backend STATIC)`.
 
-2. Build the libraries:
+2. Build the Android extension:
 
 ```bash
-cmake --build . -j16
+cmake extension/android \
+  -DCMAKE_TOOLCHAIN_FILE="${ANDROID_NDK}"/build/cmake/android.toolchain.cmake \
+  -DANDROID_ABI="${ANDROID_ABI}" \
+  -DCMAKE_INSTALL_PREFIX=cmake-android-out \
+  -Bcmake-android-out/extension/android
+
+cmake --build cmake-android-out/extension/android -j16
 ```
 
 ## Deploying on Device via Demo App
 
 ### Steps for Deploying Model via XNNPACK
 
 ```bash
-mkdir -p ../examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a
-```
-
-Copy the core libraries:
-
-```bash
-cp ./examples/demo-apps/android/jni/libexecutorch_jni.so \
-   ../examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a/libexecutorch.so
+mkdir -p examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a
+cp cmake-android-out/extension/android/libexecutorch_jni.so \
+   examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a/libexecutorch.so
 ```
 
 This allows the Android app to load ExecuTorch runtime with XNNPACK backend as a JNI library. Later, this shared library will be loaded by `NativePeer.java` in Java code.
@@ -160,15 +161,17 @@ mkdir -p ../examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64
 We need to push some additional Qualcomm HTP backend libraries to the app. Please refer to [Qualcomm docs](build-run-qualcomm-ai-engine-direct-backend.md) here.
 
 ```bash
-cp ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnHtp.so ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnHtpV69Skel.so ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnHtpStub.so ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnSystem.so \
-   ../examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a
+cp ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnHtp.so ${QNN_SDK_ROOT}/lib/hexagon-v69/unsigned/libQnnHtpV69Skel.so ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnHtpV69Stub.so ${QNN_SDK_ROOT}/lib/aarch64-android/libQnnSystem.so \
+   examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a
 ```
 
 Copy the core libraries:
 
 ```bash
-cp ./examples/demo-apps/android/jni/libexecutorch_jni.so \
-   ../examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a/libexecutorch.so
+cp cmake-android-out/extension/android/libexecutorch_jni.so \
+   examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a/libexecutorch.so
+cp cmake-android-out/lib/libqnn_executorch_backend.so \
+   examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs/arm64-v8a/libqnn_executorch_backend.so
 ```
 
 ## Running the App

examples/demo-apps/android/ExecuTorchDemo/app/build.gradle.kts

@@ -68,3 +68,12 @@ dependencies {
   debugImplementation("androidx.compose.ui:ui-tooling")
   debugImplementation("androidx.compose.ui:ui-test-manifest")
 }
+
+tasks.register("setup") {
+  doFirst {
+    exec {
+      commandLine("sh", "examples/demo-apps/android/LlamaDemo/setup.sh")
+      workingDir("../../../../../")
+    }
+  }
+}

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

@@ -1,40 +1,40 @@
 #!/usr/bin/env bash
+# Copyright (c) Meta Platforms, Inc. and affiliates.
 # All rights reserved.
 #
-# Copyright 2023-2024 Arm Limited and/or its affiliates.
-#
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree.
 
 set -eu
 
-# Note: Set up ANDROID_NDK, ANDROID_ABI, BUCK2, and FLATC
-cmake . -DCMAKE_INSTALL_PREFIX=cmake-out \
+CMAKE_OUT="${CMAKE_OUT:-cmake-out-android}"
+# Note: Set up ANDROID_NDK and ANDROID_ABI
+cmake . -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
   -DCMAKE_TOOLCHAIN_FILE="${ANDROID_NDK}/build/cmake/android.toolchain.cmake" \
   -DANDROID_ABI="${ANDROID_ABI}" \
-  -DBUCK2="${BUCK2}" \
   -DEXECUTORCH_BUILD_XNNPACK=ON \
-  -DEXECUTORCH_BUILD_FLATC=OFF \
   -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
-  -DFLATC_EXECUTABLE="${FLATC}" \
   -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-  -Bcmake-out
+  -DEXECUTORCH_BUILD_OPTIMIZED=ON \
+  -DCMAKE_BUILD_TYPE=Release \
+  -B"${CMAKE_OUT}"
 
 if [ "$(uname)" == "Darwin" ]; then
   CMAKE_JOBS=$(( $(sysctl -n hw.ncpu) - 1 ))
 else
   CMAKE_JOBS=$(( $(nproc) - 1 ))
 fi
-cmake --build cmake-out -j "${CMAKE_JOBS}" --target install
+cmake --build "${CMAKE_OUT}" -j "${CMAKE_JOBS}" --target install --config Release
 
-cmake extension/android -DBUCK2="${BUCK2}" \
+cmake extension/android \
   -DCMAKE_TOOLCHAIN_FILE=${ANDROID_NDK}/build/cmake/android.toolchain.cmake \
   -DANDROID_ABI="${ANDROID_ABI}" \
-  -DCMAKE_INSTALL_PREFIX=cmake-out \
-  -Bcmake-out/extension/android
+  -DCMAKE_INSTALL_PREFIX="${CMAKE_OUT}" \
+  -DCMAKE_BUILD_TYPE=Release \
+  -B"${CMAKE_OUT}"/extension/android
 
-cmake --build cmake-out/extension/android -j "${CMAKE_JOBS}"
+cmake --build "${CMAKE_OUT}"/extension/android -j "${CMAKE_JOBS}" --config Release
 
 JNI_LIBS_PATH="examples/demo-apps/android/ExecuTorchDemo/app/src/main/jniLibs"
 mkdir -p "${JNI_LIBS_PATH}/${ANDROID_ABI}"
-cp cmake-out/extension/android/libexecutorch_jni.so "${JNI_LIBS_PATH}/${ANDROID_ABI}/libexecutorch.so"
+cp "${CMAKE_OUT}"/extension/android/libexecutorch_jni.so "${JNI_LIBS_PATH}/${ANDROID_ABI}/"

extension/android/CMakeLists.txt

@@ -58,14 +58,9 @@ if(EXECUTORCH_BUILD_LLAMA_JNI)
   add_library(llama_runner STATIC IMPORTED)
   set_property(TARGET llama_runner PROPERTY IMPORTED_LOCATION ${LLAMA_RUNNER_PATH})
 
-  set(CUSTOM_OPS_LIB_PATH ${CMAKE_CURRENT_BINARY_DIR}/../../examples/models/llama2/custom_ops/libcustom_ops_lib.a)
-  add_library(custom_ops_lib STATIC IMPORTED)
-  set_property(TARGET custom_ops_lib PROPERTY IMPORTED_LOCATION ${CUSTOM_OPS_LIB_PATH})
-
   set(CUSTOM_OPS_PATH ${CMAKE_CURRENT_BINARY_DIR}/../../examples/models/llama2/custom_ops/libcustom_ops.a)
   add_library(custom_ops STATIC IMPORTED)
   set_property(TARGET custom_ops PROPERTY IMPORTED_LOCATION ${CUSTOM_OPS_PATH})
-  target_link_options_shared_lib(custom_ops_lib)
 
   if(TARGET pthreadpool)
     set(LLAMA_JNI_SRCS jni/jni_layer_llama.cpp ../../backends/xnnpack/threadpool/cpuinfo_utils.cpp)
@@ -82,6 +77,6 @@ if(EXECUTORCH_BUILD_LLAMA_JNI)
   endif()
   target_include_directories(executorch_llama_jni PRIVATE ${_common_include_directories})
   target_link_libraries(executorch_llama_jni ${link_libraries} llama_runner
-                        custom_ops custom_ops_lib cpublas eigen_blas)
+                        custom_ops cpublas eigen_blas)
   target_compile_options(executorch_llama_jni PUBLIC ${_common_compile_options})
 endif()