Novel Graph Processor Architecture, Prototype System, and Results
Abstract
Graph algorithms are increasingly used in applications that exploit large databases. However, conventional processor architectures are inadequate for handling the throughput and memory requirements of graph computation. Lincoln Laboratory's graph-processor architecture represents a rethinking of parallel architectures for graph problems. Our processor utilizes innovations that include a sparse matrix-based graph instruction set, a cacheless memory system, accelerator-based architecture, a systolic sorter, high-bandwidth multi-dimensional toroidal communication network, and randomized communications. A field-programmable gate array (FPGA) prototype of the new graph processor has been developed with significant performance enhancement over conventional processors in graph computational throughput.
Cite
@article{arxiv.1607.06541,
title = {Novel Graph Processor Architecture, Prototype System, and Results},
author = {William S. Song and Vitaliy Gleyzer and Alexei Lomakin and Jeremy Kepner},
journal= {arXiv preprint arXiv:1607.06541},
year = {2016}
}
Comments
7 pages, 8 figures, IEEE HPEC 2016