Yet shorter warmup by combining no-state-loss and MRRL for sampled LRU cache simulation

Sampling is a well known technique for speeding up time-consuming architectural simulations. An important issue with sampling is the hardware state at the beginning of each sampling unit. This paper presents a highly accurate and highly efficient warmup method for sampled cache simulation by combining ‘no-state-loss (NSL)’ and ‘memory reference reuse latency (MRRL)’. Our combined warmup scheme MRRL–NSL achieves the same accuracy for sampled LRU cache simulation as MRRL with a two orders of magnitude shorter warmup. Compared to NSL, MRRL–NSL has a factor 2–6 shorter warmup while inducing a small absolute miss rate error of 0.1%.