high performance computing on graphics processing units: hgpu.org

Parallelization of scientific problems is a challenging task which has a wide application area both on distributed programming, cloud computing and recently on GPGPU. Spectral graph partitioning is a widely used technique in many fields such as image processing, scientific computing, machine learning etc. In this study we analyze spectral graph partitioning subroutines on a GPGPU framework. Each step is analyzed with differing techniques to lead a conclusion about usage of GPGPU on overall spectral graph partitioning algorithms.