Evolutionary Benchmark Subsetting
Zhanpeng Jin, Allen C. Cheng · IEEE Micro · 2008
To improve simulation efficiency and relieve burdened benchmarking efforts, this research proposes a survival-of-the-fittest evolutionary methodology. The goal is to subset any given benchmark suite based on its inherent workload characteristics, desired workload space coverage, and total execution time. Given a user-specified workload space coverage threshold, the proposed technique can systematically yield the "fittest" time-efficient benchmark subset.