We examine multithreading to improve uniprocessor cost/performance on multiple processes. Processor utilization and cache behavior are studied analytically and under simulation by interleaving reference traces to model timesharing and multithreading. Multithreading a small number of threads is superior with large on-chip caches and significant memory latency. The switch need not be extremely fast. Surprisingly, miss ratios under multithreading may be lower than under timesharing, because switch on-miss multithreading favors processes with better cache behavior.
展开▼