A simulation based approach to the problem of performance evaluation o
f large switching fabrics is presented. Experimental results in the fo
rm of cell queuing and buffer occupancy distributions for switches wit
h up to 2048 inputs and outputs operating under a wide variety of load
s are presented. The traditional dilemma between simulation model vali
dity and its run-time where a more general purpose model inevitably re
sults in longer run-times, is eliminated by the introduction of a new
parameter: the architecture of the simulation platform. With model val
idity as an invariant, the simulation platform can range from a sequen
tial computer via shared memory and distributed memory general purpose
parallel processors up to a dedicated emulator. An analytical model o
f the class of applications under study gives insight in the influence
of machine and problem parameters on simulator run-time performance.
Examples of its use for the choice of an optimal parallel architecture
for a given problem and of its use for the estimation of application
behavior on a given parallel machine architecture are presented.