Publications of Torsten Hoefler

Maciej Besta, Marc Fischer, Tal Ben-Nun, Johannes de Fine Licht, Torsten Hoefler:

Substream-Centric Maximum Matchings on FPGA

(Feb. 2019, In Proceedings of the 27th ACM/SIGDA International Symposium on Field-Programmable Gate Arrays )
Best Paper Finalist (4/30)

Abstract

Developing high-performance and energy-efficient algorithms for maximum matchings is becoming increasingly important in social network analysis, computational sciences, scheduling, and others. In this work, we propose the first maximum matching algorithm designed for FPGAs; it is energy-efficient and has provable guarantees on accuracy, performance, and storage utilization. To achieve this, we forego popular graph processing paradigms, such as vertex-centric programming, that are tuned for CPUs and often entail large communication costs. Instead, we propose a substream-centric approach, in which the input stream of data is divided into substreams processed independently to enable more parallelism while lowering communication costs. We base our work on the theory of streaming graph algorithms and analyze 15 models and 28 algorithms. We use this analysis to provide theoretical underpinning that matches well the physical constraints of FPGA platforms. Our algorithm delivers high performance (more than 4× speedup over tuned parallel CPU variants), low memory, high accuracy, and effective usage of FPGA resources. The substream-centric approach could easily be extended to other algorithms to offer low-power and high-performance graph processing on FPGAs.

Documents

download article:

download slides:

BibTeX

@inproceedings{,
  author={Maciej Besta and Marc Fischer and Tal Ben-Nun and Johannes de Fine Licht and Torsten Hoefler},
  title={{Substream-Centric Maximum Matchings on FPGA}},
  year={2019},
  month={Feb.},
  note={In Proceedings of the 27th ACM/SIGDA International Symposium on Field-Programmable Gate Arrays},
  source={http://www.unixer.de/~htor/publications/},
}


serving: 18.216.174.32:9470	© Torsten Hoefler