Loading [MathJax]/extensions/MathZoom.js
Martini: A Network Interface Controller Chip for High Performance Computing with Distributed PCs | IEEE Journals & Magazine | IEEE Xplore

Martini: A Network Interface Controller Chip for High Performance Computing with Distributed PCs


Abstract:

In this paper, “Martini,” a network interface controller chip for our original network called RHiNET is described. Martini is designed to provide high-bandwidth and low-l...Show More

Abstract:

In this paper, “Martini,” a network interface controller chip for our original network called RHiNET is described. Martini is designed to provide high-bandwidth and low-latency communication with small overhead. To obtain high performance communication, protected user-level zero-copy RDMA communication functions are completely implemented by a hardwired logic. Also, to reduce the communication latency efficiently, we have proposed PIO-based communication mechanisms called “On-the-fly (OTF)” and have implemented them on Martini. The evaluation results show that Martini connected to a 64bit/66MHz PCI-bus achieves 470MByte/s maximum bidirectional bandwidth and 1.74 μsec minimum latency on host-to-host memory copying.
Published in: IEEE Transactions on Parallel and Distributed Systems ( Volume: 18, Issue: 9, September 2007)
Page(s): 1282 - 1295
Date of Publication: 13 August 2007

ISSN Information:


1 Introduction

Today, the performance improvement and cost reduction of personal computers are remarkable. Many companies and education facilities have been introducing a lot of PCs as powerful and economical computational resources. However, most of them are used only for lightweight tasks such as documentation, e-mail, Web browsing, and so on. Furthermore, they are not used during nights and holidays. Thus, it can be considered that they have a large amount of unused computational power, and the total of them becomes enormous. Grid [1] and some other coarse-grain distributed computing are ideas to utilize surplus computational power. SETI@home [2] is one of the well-known succeeding applications of the coarse-grain distributed computing.

References

References is not available for this document.