GNNBuilder#

GNNBuilder is a framework for generating FPGA hardware accelerators for graph neural networks (GNNs) using High Level Synthesis (HLS).

GNNBuilder is developed and maintained by the Stefan Abi-Karam from Sharc Lab at Georgia Tech.

Quick Guide#

See the sidebar for links to diffetent secitons of the documentation.

  • Framework Overview - Overview of the GNNBuilder framework explaining the different components and how they work together.

  • Setup - Instructions for installing and setting up GNNBuilder.

  • Simple Tutorial - A simple tutorial for using GNNBuilder to generate a complete GNN accelerator from start to finish incluidng testbench generation, testbench evaluation, HLS Synthesis, IP export for Vivado, IP export for Vitis, bitsretam generation for the Vitis flow, and on-device execution of the GNN acclerator using the Vitis flow and XRT runtime.

Publications and Source Code#

GNNBuilder has been published in the following places:

The source code repository is hosted on GitHub:

Citing and Referencing#

If you use GNNBuilder in your research, please cite the primary FPL 2023 confrence paper:

S. Abi-Karam and C. Hao, "GNNBuilder: An Automated Framework for Generic Graph Neural Network Accelerator Generation, Simulation, and Optimization," in 2023 33nd International Conference on Field-Programmable Logic and Applications (FPL), Gothenburg, Sweden: IEEE, Sep. 2023.

@inproceedings{abi-karam_gnnbuilder_2023,
   location = {Gothenburg, Sweden},
   title = {{GNNBuilder}: An Automated Framework for Generic Graph Neural Network Accelerator Generation, Simulation, and Optimization},
   eventtitle = {2023 33nd International Conference on Field-Programmable Logic and Applications ({FPL})},
   booktitle = {2023 33nd International Conference on Field-Programmable Logic and Applications ({FPL})},
   publisher = {{IEEE}},
   author = {Abi-Karam, Stefan and Hao, Cong},
   date = {2023-09},
}