feat(imperative/opr): fix extern c opr mace example md

GitOrigin-RevId: b83fa4568b4659e539c827378f9a02e5161ea1af

sdk/c-opr-loaders/mace/Makefile

@@ -15,9 +15,10 @@ endif
 TARGET := libmace_loader.so
 all: $(TARGET)
 
-CXX = aarch64-linux-android-g++
+# can be find at NDK_ROOT/toolchains/llvm/prebuilt/linux-x86_64/bin/
+CXX = aarch64-linux-android21-clang++
 
-CXXFLAGS = -std=c++14  -I$(SDKPATH)/include \
+CXXFLAGS = -std=c++14 -I$(SDKPATH)/include \
 			-Dmgb_mace_loader_init=mgb_c_opr_init
 
 DEBUG ?= 0
@@ -27,12 +28,7 @@ else
     CXXFLAGS += -O2
 endif
 
-LDFLAGS = -L$(SDKPATH)/lib \
-			-lmace_static -lcore -lgenerated_version \
-			-lport_android -lport_linux_base -lport_posix -lport_base \
-			-lproto -lutils -lgenerated_opencl_kernel \
-			-lops -lcore -lprotobuf-lite -llog \
-			-lpthread -l:rpcmem.a
+LDFLAGS = -L$(SDKPATH)/lib -lmace -llog
 
 LDLIBS = -shared -fPIC
 

sdk/c-opr-loaders/mace/README.md

 # Extern-C-Opr with MACE
 
 ### Build MegEngine `load_and_run` for arm64-v8a
+NOTICE: build depends on [NDK](https://developer.android.com/ndk/downloads)
+after download, please config env by:
+```bash
+export NDK_ROOT=path/to/ndk
+export ANDROID_NDK_HOME=${NDK_ROOT}
+export PATH=${NDK_ROOT}/toolchains/llvm/prebuilt/linux-x86_64/bin/:$PATH
+```
 
 ```bash
 cd $MEGENGINE_HOME
+git checkout v1.0.0    (we only test v1.0.0 version)
 ./scripts/cmake-build/cross_build_android_arm_inference.sh -a arm64-v8a -r
 ```
 
-After successfully built, load_and_run should be in `$MEGENGINE_HOME/build_dir/android/arm64-v8a/Release/install/bin`
+After successfully built:
+* load_and_run should be in `$MEGENGINE_HOME/build_dir/android/arm64-v8a/Release/install/bin`
+* libmegengine.so should be in `$MEGENGINE_HOME/build_dir/android/arm64-v8a/Release/install/lib`
 
 ### Build MACE libraries for arm64-v8a with GPU runtime
 
 ```bash
 cd $MACE_HOME
 RUNTIME=GPU bash tools/cmake/cmake-build-arm64-v8a.sh
-
-cp -r $MACE_HOME/build/cmake-build/arm64-v8a/install $MEGENGINE_HOME/sdk/c-opr-loaders/mace/arm64-v8a
+export SDKPATH=${MACE_HOME}/build/cmake-build/arm64-v8a/install
 ```
+After successfully libmace.so should be in `$MACE_HOME/build/cmake-build/arm64-v8a/install/lib/libmace.so`
 
 ### Build MACE loader for MegEngine
 
-```
-SDK_PATH=/path/to/mace-sdk make
-```
-
-If `SDK_PATH` is not set, by default it's `./arm64-v8a`
+If `SDKPATH` is not set, by default it's `./arm64-v8a`
 
 You can run with debug mode(by adding `DEBUG=1` to make command), which will show more running information
 
 ### Prepare a MACE model(for example: resnet_50), wrap it with MegEngine extern c opr
 
 ```
-python3 dump_model.py path/to/resnet_50.pb path/to/resnet_50.data path/to/resnet_50.mdl path/to/resnet_50.yml
+python3 dump_model.py --input path/to/resnet_50.pb --param path/to/resnet_50.data --output resnet_50.mdl --config path/to/resnet_50.yml
 ```
 
 `*.pb` file denotes the model structure, `*.data` denotes the model parameters
@@ -40,14 +46,22 @@ Check [here](https://github.com/XiaoMi/mace-models) to learn how to write yml fi
 
 ### Run with load-and-run
 
-First of all, send all files to the executed device:
+First of all, send all files to the executed device(for example: /data/local/tmp/test/):
 
 - load_and_run
 - resnet_50.mdl
 - libmace_loader.so
+- libmegengine.so
+- libmace.so
+
+As mace build with `c++_shared` by default, but old AOSP device do not have `libc++_shared.so` by default, if you use this class devices
+also need send it to devices, which always can be found at `${NDK_ROOT}/sources/cxx-stl/llvm-libc++/libs/arm64-v8a/libc++_shared.so`
 
 ```
-MGB_MACE_RUNTIME=GPU MGB_MACE_OPENCL_CACHE_PATH=/path/to/opencl MGB_MACE_LOADER_FORMAT=NCHW /path/to/load_and_run /path/to/resnet_50.mdl --c-opr-lib /path/to/libmace_loader.so
+login to device
+cd /path/to/ (for example: /data/local/tmp/test/)
+
+MGB_MACE_RUNTIME=GPU MGB_MACE_OPENCL_CACHE_PATH=./ MGB_MACE_LOADER_FORMAT=NCHW LD_LIBRARY_PATH=. ./load_and_run resnet_50.mdl  --c-opr-lib libmace_loader.so  --input input-bs1.npy
 ```
 
 RUNTIME candidates:
@@ -77,4 +91,4 @@ To generate the tunig param file, give `MACE_TUNING=1` env and set the `MACE_RUN
 
  # then run test using the param
  MGB_MACE_TUNING_PARAM_PATH=opencl/vgg16.tune_param MGB_MACE_RUNTIME=GPU MGB_MACE_OPENCL_PATH=opencl MGB_MACE_LOADER_FORMAT=NCHW ./load_and_run mace/vgg16.mdl --c-opr-lib libmace_loader.so --input 4d.npy
- ```
\ No newline at end of file
+ ```