Threads vs. caches: Modeling the behavior of parallel workloads | IEEE Conference Publication | IEEE Xplore