update README.md file

- updated README.md file according to review by Executorch team

Signed-off-by: jiseong.oh <[email protected]>

Author: Jiseong-oh

backends/samsung/README.md

@@ -1,9 +1,9 @@
 # ExecuTorch Samsung Exynos Delegate
 
-The subtree contains Exynos delegation implementation for ExecuTorch. The target of delegation
-is deploying torch model run with exynos NPU/DSP.
+The subtree contains Exynos delegate implementation for ExecuTorch. The target of delegation
+is deploying torch model to run with exynos NPU/DSP.
 
-This backend is implemented on the top of [EXYNOS_LITECORE](https://soc-developer.semiconductor.samsung.com/global/development/light-core)
+This backend is implemented on the top of [EXYNOS_LITECORE](https://soc-developer.semiconductor.samsung.com/global/development/ai-litecore)
 Please prepare the SDK before you start, it is important to code compilation and runtime.
 
 ## Delegate Options
@@ -33,26 +33,26 @@ examples
 ```
 
 ## How to build
-Please download Exynos AI LiteCore, and set the root path of SDK directory to `EXYNOS_AI_LITECORE_PATH`.</br>
+Please download Exynos AI LiteCore, and set the root path of SDK directory to `EXYNOS_AI_LITECORE_ROOT`.</br>
 Please navigate to [Android NDK](https://developer.android.com/ndk) and download a version of NDK.
 `ANDROID_NDK` refers the root path of NDK directory.</br>
 
 ### Set up environment variables
 ```bash
-export LD_LIBRARY_PATH=${EXYNOS_AI_LITECORE_PATH}/lib64
+export LD_LIBRARY_PATH=${EXYNOS_AI_LITECORE_ROOT}/lib/x86_64-linux/
 ```
 
 ### Build AOT Targets
-Generated python artifacts allow user call `Compile` interface to lower a model to ENN backend in python script.
+Generates python artifacts that allow user call `Compile` interface to lower a model to Exynos backend in python script.
 ```bash
 ./backends/samsung/build.sh -b x86_64
 ```
 
-### Build ENN Runtime
+### Build ENN Executor Runner
 ```bash
 ./backends/samsung/build.sh -b android --ndk ${ANDROID_NDK}
 ```
-ANDROID_ABI=arm64-v8a is default, necessary runtime executable generate in `build_exynos_android` directory.
+ANDROID_ABI=arm64-v8a is default, necessary runtime executable generated in `build_exynos_android` directory.
 
 ### Build Anroid Extension
 This is later exposed Java app. Please turn on CMake option `EXECUTORCH_BUILD_ENN`, and ENN runtime will be added.
@@ -67,5 +67,6 @@ cmake --build cmake-android-out/extension/android -j8
 ```
 
 ## Examples
+python -m executorch.examples.samsung.aot_compiler --chipset e9955 -m ic3
 
 Please see this [README.md](../../examples/samsung/README.md).

examples/samsung/README.md

@@ -25,15 +25,17 @@ Check the chip's version, when lower the model, set the corresponding chip versi
 ## Lowering
 
 Before running an example, please copy python artifacts to target directory `PYTHON_TARGET_DIR`.
+If you use `build.sh` script to compile samsung backend, please skip the copy step.
+
 Set up `PYTHON_TARGET_DIR` to `${EXECUTORCH_ROOT}/backends/samsung/python`.
 ```bash
 cp -rf ${EXECUTORCH_ROOT}/build_x86_64/backends/samsung/Py*.so ${PYTHON_TARGET_DIR}
 cp -rf ${EXYNOS_AI_LITECORE_PATH}/python/snc_py_api*.so ${PYTHON_TARGET_DIR}
 ```
 
-Take `EXECUTORCH_ROOT` as work directory and here is an example for mul.
+Take `EXECUTORCH_ROOT` as work directory and here is an example for ic3.
 ```bash
-python -m executorch.examples.samsung.aot_compiler -m mul
+python -m executorch.examples.samsung.aot_compiler --chipset E9955 -m ic3 --output_dir ic3_artifact
 ```
 
 ## Execution
@@ -45,15 +47,14 @@ After lowering, we could get a pte model and then run it on mobile phone.
 DEVICE_DIR=/data/local/tmp/executorch
 adb shell mkdir ${DEVICE_DIR}
 adb push ${EXECUTORCH_ROOT}/cmake-android-out/backends/samsung/enn_executor_runner ${DEVICE_DIR}
-adb push ${EXYNOS_AI_LITECORE_PATH}/enn_api/libenn* ${DEVICE_DIR}
 ```
 
 #### Step 2: Indicate dynamic linkers and execute model
 ```bash
-adb push ./mul_exynos_fp32.pte ${DEVICE_DIR}
+adb push ./ic3_exynos_fp32.pte ${DEVICE_DIR}
 adb shell "cd ${DEVICE_DIR} \
            && export LD_LIBRARY_PATH=${DEVICE_DIR} \
-           && ./enn_executor_runner -model ./mul_exynos_fp32.pte"
+           && ./enn_executor_runner -model ./ic3_exynos_fp32.pte -input ./ic3_input_0.bin --output_path ."
 ```
 
 `enn_executor_runner` has more usages, please refer to the help message.