Elsevier

Computer Communications

Volume 20, Issue 12, 1 November 1997, Pages 1089-1097
Computer Communications

Interconnection network front-end controller combining to reduce hot spots effects

https://doi.org/10.1016/S0140-3664(97)00093-5Get rights and content

Abstract

Multistage interconnection networks are very promising for shared-memory multiprocessor systems. These networks offer flexibility, scalability, and good performance-cost ratio. However, under a non-uniform traffic pattern, the performance of multistage interconnection networks suffers greatly because of hot spot traffic contention. The potential performance degradation due to even moderate hot spot traffic was found to be very significant, severely reducing all memory access, not just to shared locations [1]. Several techniques have been proposed to reduce the effects of hot spot and tree saturation contentions in multistage interconnection networks. These include multibuffered switching nodes, multipath networks, and request combining [2]. Request combining strategy was found to be an effective method of reducing the tree saturation problem [1]. Some request combining approaches have been proposed [1,3–5].

This paper proposes a new request combining based architecture to reduce the hot spot performance degradation. This approach is referred to as interconnection network front-end controller combining (IN-FEC). In IN-FEC, the interconnection network combines requests to the shared memory location, and the memory front-end controllers are used to decombine the combined request and distribute the results. Simulation results show considerable throughput enhancement under IN-FEC for an Omega network.

References (15)

There are more references available in the full text version of this article.

Cited by (3)

  • Switch fabric design for high performance IP routers: A survey

    2005, Journal of Systems Architecture
    Citation Excerpt :

    However, buffered Banyan/Delta networks are very sensitive to non-uniform traffics, especially hot spot, regardless of the buffering strategy, due to a phenomenon called “tree-saturation effect” which can severely degrade the performance of all traffic within a MIN [61,70,121] under hot-spot traffic patterns. Example solutions to improve throughput performance in hot-spot situations are: (1) extra stages of switching elements to provide alternate paths [138]; (2) networks using combining techniques or with combining switches [44,102,120,134]; (3) use of multiple parallel networks [57]; and (4) use of flow-control techniques to avoid congestion [27,47,113]. Items 1 and 3 are self-explanatory.

  • Switch fabric architecture analysis for a scalable bi-directionally reconfigurable IP router

    2004, Journal of Systems Architecture
    Citation Excerpt :

    Numerous solutions to improve throughput performance in hot-spot are available. Examples are (1) extra stages of switch elements to provide alternate paths [50]; (2) networks using combining techniques or with combining switches [15,41,49]; (3) use of multiple parallel networks [18]; and (4) use of flow-control techniques to avoid congestion [10,16,38]. However, major drawbacks are the need for intense HW logic making it extremely expensive, and for the cases of combining networks and of use of flow-control based techniques, out-of-order packet delivery could result [56].

View full text