Bandwidth Enhancement between Graphics Processing Units on the Peripheral Component Interconnect Bus

Abstract

General purpose computing on graphics processing units is a new trend in high performance computing. Present day applications require office and personal supercomputers which are mostly based on many core hardware accelerators communicating with the host system through the Peripheral Component Interconnect (PCI) bus. Parallel data compression is a difficult topic but compression has been used successfully to improve the communication between parallel message passing interface (MPI) processes on high performance computing clusters. In this paper we show that special pur pose compression algorithms designed for scientific floating point data can be used to enhance the bandwidth between 2 graphics processing unit (GPU) devices on the PCI Express (PCIe) 3.0 x16 bus in a homebuilt personal supercomputer (PSC).

Keywords

• Bandwidth
• Compression
• GPU
• PCI