Application-aware Prefetch Prioritization in On-chip Networks

Data prefetching is an eective technique for hiding memory la-tency. When issued prefetches are inaccurate, performance can degrade. Prior research provided solutions to deal with inaccu-rate prefetches at the cache and memory levels, but not in the in-terconnect of a large-scale multiprocessor system. This work in-troduces application-aware prefetch prioritization techniques to mitigate the negative eects of prefetching in a network-on-chip (NoC) based multicore system. The idea is to rank prefetches from dierent applications based on their potential utility for the application and propensity to cause interference to other ap-plications. Our evaluation shows that this approach provides signi cant performance improvements over a baseline that does not distinguish between prefetches from dierent applications