arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1607.06541cs.DCcs.AR
收藏

新型图形处理器体系结构、原型系统和结果

Novel Graph Processor Architecture, Prototype System, and Results

William S. Song, Vitaliy Gleyzer, Alexei Lomakin, Jeremy Kepner

摘要:

图算法越来越多地用于利用大型数据库的应用程序中。然而,传统的处理器架构不足以处理图形计算的吞吐量和存储器要求。林肯实验室的图形处理器架构代表了对图形问题的并行架构的重新思考。我们的处理器采用的创新,包括稀疏矩阵为基础的图形指令集,无缓存存储器系统,基于加速器的架构,脉动排序器,高带宽多维环形通信网络,和随机通信。一个现场可编程门阵列(FPGA)原型的新的图形处理器已开发出显着的性能增强,在传统的处理器图形计算吞吐量。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑