Dynamic Access Ordering: Bounds on Memory BandwidthReport
Memory bandwidth is becoming the limiting performance factor for many applications, particu- larly scientific computations. Access ordering is one technique that can help bridge the processor- memory performance gap. We are part of a team developing a combined hardware/software scheme for implementing access ordering dynamically at run-time. The hardware part of this solution is the Stream Memory Controller, or SMC. In order to validate the SMC concept, we have conducted numerous simulation experiments, the results of which are presented elsewhere. We have developed analytical models to bound asymptotic uniprocessor SMC performance, and have demonstrated that the simulation behavior of our dynamic access-ordering heuristics approaches those bounds. Here we introduce a model of of SMC startup costs, and we extend the uniprocessor SMC models to describe performance for modest-sized symmetric multiprocessor (SMP) SMC systems.
All rights reserved (no additional license for public reuse)
McKee, Sally. "Dynamic Access Ordering: Bounds on Memory Bandwidth." University of Virginia Dept. of Computer Science Tech Report (1994).
University of Virginia, Department of Computer Science