Abstract

This paper compares the trace-sampling techniques of set sampling and time sampling. Using the multi-billion-reference traces of Borg et al., we apply both techniques to multi-megabyte caches, where sampling is most valuable. We evaluate whether either technique meets a 10% sampling goal: using 10% of the references in a trace can it estimate the trace's true misses per instruction with 10% relative error and at least 90% confidence. Our results show that set sampling meets the 10% sampling goal, while time sampling does not. We also find that cold-start bias in time samples is most effectively reduced by the technique of Wood et al. Nevertheless, overcoming cold-start bias requires the use of tens of millions consecutive references. Index Terms - Cache memory, cache performance, cold start, computer architecture, memory systems, performance evaluation, sampling techniques, trace-driven simulation. 1. Introduction Computer designers commonly use trace-driven simulation to evaluate al...

Citations