Presentation 2009-01-13
Evalutions of Prediction Router for Low-Latency On-Chip Networks
Hiroki MATSUTANI, Michihiro KOIBUCHI, Hideharu AMANO, Tsutomu YOSHINAGA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) To reduce the communication latency on recent many-core architectures, we have proposed a low-latency router architecture that predicts an output channel being used by the next packet transfer and speculatively completes the switch arbitration. In the prediction routers, incoming packets are transferred without waiting the routing computation and switch arbitration if the prediction hits. Thus, the primary concern for reducing the communication latency is to select the prediction algorithm that offers a high hit rate for a given network environment. In this paper, the prediction routers support multiple prediction algorithms and select one of them in response to the network topology, routing algorithm, and traffic pattern. To investigate optimal prediction algorithm for a given network, the prediction router architecture is applied to four many-core architectures. For each case study, the router is designed with a 65nm CMOS process and evaluated in terms of the area, energy, prediction hit rate, and latency. This paper shows their area- and energy-overhead, and the pros and cons of each prediction algorithm.
Keyword(in Japanese) (See Japanese page)
Keyword(in English)
Paper # ICD2008-129
Date of Issue

Conference Information
Committee ICD
Conference Date 2009/1/6(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Integrated Circuits and Devices (ICD)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Evalutions of Prediction Router for Low-Latency On-Chip Networks
Sub Title (in English)
Keyword(1)
1st Author's Name Hiroki MATSUTANI
1st Author's Affiliation Graduate School of Science and Technology, Keio University()
2nd Author's Name Michihiro KOIBUCHI
2nd Author's Affiliation National Institute of Informatics
3rd Author's Name Hideharu AMANO
3rd Author's Affiliation Graduate School of Science and Technology, Keio University
4th Author's Name Tsutomu YOSHINAGA
4th Author's Affiliation Graduate School of Information Systems, The University of Electro-Communications
Date 2009-01-13
Paper # ICD2008-129
Volume (vol) vol.108
Number (no) 375
Page pp.pp.-
#Pages 6
Date of Issue