/TLPGNN

Source codes for TLPGNN paper published at HPDC'22

Primary LanguageCuda

Software requirement

gcc

scipy.sparse

CUDA (11.3 tested)

Pytorch (1.10 tested)

Hardware

GPU: V100 (tested)

Execute

cd gcn/
python test_kernel.py --dataset citeseer --size 32

To run TLPGNN on other dataset, please add the data folder under data/ and provide the same data files with our citeseer example.

Reference

If you use TLPGNN in your project, please cite the following paper.

@inproceedings{fu2022tlpgnn,
  title={TLPGNN: A Lightweight Two-Level Parallelism Paradigm for Graph Neural Network Computation on GPU},
  author={Fu, Qiang and Ji, Yuede and Huang, H Howie},
  booktitle={Proceedings of the 31st International Symposium on High-Performance Parallel and Distributed Computing},
  pages={122--134},
  year={2022}
}