Skip to content

Latest commit

 

History

History

genesys

README.md

Welcome to GeneSys!

The GeneSys compiler uses an embedded architecture description language to create a Hierarchical Architecture Graph for flexibly compiling mg-DFGs to different architectures. The GeneSys architecture uses a systolic array centric approach that forms the core convolution engine for implementing DNN algorithms, and can be customized to run a range of standard DNN topologies.

This document will help you get up and running.

Step 0: Check prerequisites

The following dependencies must be met by your system:

  • python >= 3.7 (For PEP 560 support)

Step 1: Clone the VeriGOOD-ML source code

$ git clone --recurse-submodules https://github.com/VeriGOOD-ML/public
$ cd public

Step 2: Create a Python virtualenv

Note: You may choose to skip this step if you are doing a system-wide install for multiple users. Please DO NOT skip this step if you are installing for personal use and/or you are a developer.

$ python -m venv ~/.venv/verigood_ml
$ source ~/.venv/verigood_ml/bin/activate
$ python -m pip install pip setuptools wheel --upgrade

Step 3: Install PolyMath and the Codelets Compiler

If you already have a working installation of Python 3.7 or Python 3.8, the easiest way to install PolyMath and the Codelets compiler is:

$ cd polymath && python -m pip install -r requirements.txt && python setup.py build install
$ cd ../codelets.src && python -m pip install -r requirements.txt && python setup.py build install && cd ..

Step 4: Compile a benchmark using GeneSys

You can compile a GeneSys benchmark by running the following commands, where <model_name> is one of resnet18, or resnet50 (you can try compiling with other onnx models you find as well!) and <config_name> is the name of one of the configs located in genesys/examples/genesys/configs:

$ python compile_benchmark.py --model <model_name> --config <config_file>

Compiled output will be stored in the compilation_output/ directory.