1# Build Instructions 2 3This document provides a step-by-step guide to set up the build environment for the MediaTek ExercuTorch libraries. 4 5## Prerequisites 6 7Before you begin, ensure you have the following prerequisites installed and configured: 8 9### 1. Buck2 Build Tool 10 11- **Download Buck2**: Obtain Buck2 from the official [releases page](https://github.com/facebook/buck2/releases/tag/2024-02-01). 12- **Add to PATH**: Extract the downloaded file and add the directory to your system's `$PATH` environment variable. 13 ```bash 14 export PATH=<path_to_buck>:$PATH 15 ``` 16 17### 2. Android NDK 18 19- **Download Android NDK**: Acquire the Android NDK version 26.3.11579264 from the [Android developer site](https://developer.android.com/ndk/downloads). 20- **Set NDK Path**: Ensure that the `$ANDROID_NDK` environment variable is set to the path where the NDK is located. 21 ```bash 22 export ANDROID_NDK=<path_to_android_ndk> 23 ``` 24 25### 3. MediaTek ExercuTorch Libraries 26 27Download [NeuroPilot Express SDK](https://neuropilot.mediatek.com/resources/public/npexpress/en/docs/npexpress) from MediaTek's NeuroPilot portal: 28 29- `libneuronusdk_adapter.mtk.so`: This universal SDK contains the implementation required for executing target-dependent code on the MediaTek chip. 30- `libneuron_buffer_allocator.so`: This utility library is designed for allocating DMA buffers necessary for model inference. 31- `mtk_converter-8.8.0.dev20240723+public.d1467db9-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl`: This library preprocess the model into a MediaTek representation. 32- `mtk_neuron-8.2.2-py3-none-linux_x86_64.whl`: This library converts the model to binaries. 33 34## Setup 35 36Follow the steps below to setup your build environment: 37 381. **Setup ExercuTorch Environment**: Refer to the [Setting up ExercuTorch](https://pytorch.org/executorch/stable/getting-started-setup) guide for detailed instructions on setting up the ExercuTorch environment. 39 402. **Setup MediaTek Backend Environment** 41- Install the dependent libs. Ensure that you are inside backends/mediatek/ directory 42 ```bash 43 pip3 install -r requirements.txt 44 ``` 45- Install the two .whl downloaded from NeuroPilot Portal 46 ```bash 47 pip3 install mtk_neuron-8.2.2-py3-none-linux_x86_64.whl 48 pip3 install mtk_converter-8.8.0.dev20240723+public.d1467db9-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl 49 ``` 50- Set evironment variables for building backend 51 ```bash 52 export NEURON_BUFFER_ALLOCATOR_LIB=<path_to_buffer_allocator> 53 ``` 54 55## Build 56 571. **Build MediaTek Backend**: Once the prerequisites are in place, run the `mtk_build.sh` script to start the build process, MediaTek backend will be built under `cmake-android-out/backends/` as `libneuron_backend.so` 58 59 ```bash 60 ./mtk_build.sh 61 ``` 62 63## Run 64 651. **Push MediaTek universal SDK and MediaTek backend to the device**: push `libneuronusdk_adapter.mtk.so` and `libneuron_backend.so` to the phone and export it to the `$LD_LIBRARY_PATH` environment variable before executing ExercuTorch with MediaTek backend. 66 67 ```bash 68 export LD_LIBRARY_PATH=<path_to_usdk>:<path_to_neuron_backend>:$LD_LIBRARY_PATH 69 ``` 70