This is the source code repo for the original work "gHyPart: GPU-friendly End-to-End Hypergraph Partitioner" accepted by ACM TACO.
The benchmark set used in our work is available on zenodo and can also be downloaded from here.
git clone --recurse-submodules https://github.com/zlwu92/gHyPart_TACO.git
cd cpu_works/BiPart/
and cd cpu_works/Mt-KaHyPar-SDet/
mkdir build && cd build/
cmake .. -DCMAKE_BUILD_TYPE=RELEASE
make bipart-cpu -j
and make MtKaHyPar -j
Go to gHyPart's root dir.
mkdir build && cd build/ && cmake ..
make -j
Go to scripts/
folder and use python run-xxx.py
to run all experiments.
Go to figure_plotting
folder and run all scripts, you can get the figures plotted in paper.