论文信息 - Worst-case Traffic for Oblivious Routing Functions

Worst-case Traffic for Oblivious Routing Functions

This paper presents an algorithm to find a worst-case trafficpattern for any oblivious routing algorithm on an arbitrary interconnectionnetwork topology. The linearity of channel loading offered by obliviousrouting algorithms enables the problem to be mapped to a bipartitemaximum-weight matching, which can be solved in polynomial time forrouting functions with a polynomial number of paths. Finding exact worstcaseperformance was previously intractable, and we demonstrate an examplecase where traditional characterization techniques overestimate thethroughput of a particular routing algorithm by 47%.

William J. Dally | Brian Towles

[1] Bruce M. Maggs,et al. Randomized Routing and Sorting on Fixed-Connection Networks , 1994, J. Algorithms.

[2] Lawrence Snyder,et al. The Case for Chaotic Adaptive Routing , 1997, IEEE Trans. Computers.

[3] Bruce M. Maggs,et al. Fast algorithms for bit-serial routing on a hypercube , 1990, SPAA '90.

[4] Sudhakar Yalamanchili,et al. Interconnection Networks: An Engineering Approach , 2002 .

[5] Allan Borodin,et al. Routing, Merging, and Sorting on Parallel Models of Computation , 1985, J. Comput. Syst. Sci..

[6] Christos Kaklamanis,et al. Tight bounds for oblivious routing in the hypercube , 1990, SPAA '90.

[7] S. Lennart Johnsson,et al. ROMM routing on mesh and torus networks , 1995, SPAA '95.

[8] H. Kuhn. The Hungarian method for the assignment problem , 1955 .

[9] William J. Dally,et al. A delay model and speculative architecture for pipelined routers , 2001, Proceedings HPCA Seventh International Symposium on High-Performance Computer Architecture.

[10] Baruch Awerbuch,et al. Universal-stability results and performance bounds for greedy contention-resolution protocols , 2001, JACM.

[11] Brian Towles. Finding Worst-case Permutations for Oblivious Routing Algorithms , 2001 .