| Performance evaluation of a new routing strategy for irregular networks with source routing |
| Full text |
Pdf
(860 KB)
|
| Source
|
International Conference on Supercomputing
archive
Proceedings of the 14th international conference on Supercomputing
table of contents
Santa Fe, New Mexico, United States
Pages: 34 - 43
Year of Publication: 2000
ISBN:1-58113-270-0
|
|
Authors
|
|
J. Flich
|
Dpto. Informática de Sistemas y Computadores, Universidad Politécnica de Valencia, Camino de Vera, 14, 46071-Valencia, Spain
|
|
M. P. Malumbres
|
Dpto. Informática de Sistemas y Computadores, Universidad Politécnica de Valencia, Camino de Vera, 14, 46071-Valencia, Spain
|
|
P. López
|
Dpto. Informática de Sistemas y Computadores, Universidad Politécnica de Valencia, Camino de Vera, 14, 46071-Valencia, Spain
|
|
J. Duato
|
Dpto. Informática de Sistemas y Computadores, Universidad Politécnica de Valencia, Camino de Vera, 14, 46071-Valencia, Spain
|
|
| Sponsor |
|
| Publisher |
|
| Bibliometrics |
Downloads (6 Weeks): 4, Downloads (12 Months): 20, Citation Count: 2
|
|
|
ABSTRACT
Networks of workstations (NOWs) are becoming increasingly popular as a cost-effective alternative to parallel computers. Typically, these networks connect processors using irregular topologies, providing the wiring flexibility, scalability, and incremental expansion capability required in this environment. In some of these networks, messages are delivered using the up*/down* routing algorithm [9]. However, the up*/down* routing scheme is often non-minimal. Also, some of these networks use source routing [1]. With this technique, the entire path to destination is generated at the source host before the message is sent.In this paper we develop a new mechanism in order to improve the performance of irregular networks with source routing, increasing overall throughput. With this mechanism, messages always use minimal paths. To avoid possible deadlocks, when necessary, routes between a pair of hosts are divided into sub-routes, and a special kind of virtual cut-through is performed at some intermediate hosts. We evaluate the new mechanism by simulation using parameters taken from the Myrinet network. We show that the current routing schemes used in Myrinet can be improved by modifying only the routing software without increasing its overhead significantly and, most importantly, without modifying the network hardware. The benefits of using the new routing scheme are noticeable for networks with 16 or more switches, and increase with network size. For 32 and 64-switch networks, throughput is increased on average by a factor ranging from 1.3 to 3.3.
REFERENCES
Note: OCR errors may be found in this Reference List extracted from the full text article. ACM has opted to expose the complete List rather than only correct and linked references.
| |
1
|
Nanette J. Boden , Danny Cohen , Robert E. Felderman , Alan E. Kulawik , Charles L. Seitz , Jakov N. Seizovic , Wen-King Su, Myrinet: A Gigabit-per-Second Local Area Network, IEEE Micro, v.15 n.1, p.29-36, February 1995
[doi> 10.1109/40.342015]
|
 |
2
|
|
| |
3
|
|
| |
4
|
|
| |
5
|
GM protocol, 'http://www.myri.com/GM'
|
| |
6
|
J. Kim and A. Chien, "An evaluation of the planar/adaptive routing," in Proc. Jth IEEE Int. Syrup. Parallel Distributeed ProcesSing, 1992
|
| |
7
|
P. R. Miller, "Efficient communications for fine-grain distributed computers," Ph.D. Thesis, Southampton University, 1991.
|
| |
8
|
Myrinet, 'M2-CB-35 LAN cables, http://www.myri.com/myrinet/product _list .html'
|
| |
9
|
M. D. Schroeder et al., "Autonst: A high-speed, self-configuring local area network using point-to-point links," Technical Report SRC research report 59, DEC, April 1990.
|
| |
10
|
|
| |
11
|
R. Sheifert, "Gigabit Ethernet," in Addison-Wesley, ISBN: 0-201-18553-9, April 1998.
|
| |
12
|
|
| |
13
|
|
|