Latency Tolerance Through Multithreading In Large-Scale Multiprocessors

In large-scale distributed-memory multiprocessors, remote memory accesses suffersignificant latencies. Caches help alleviate the memory latency problem by maintaininglocal copies of frequently used data. However, they cannot eliminate the latencycaused by first-time references and invalidations needed to enforce cache coherence.Multithreaded processors tolerate such latencies by rapidly switching between threadswhen they encounter cache misses. This paper evaluates the effectiveness...