Cache performance in vector supercomputers
Leonidas I. Kontothanassis, Rabin A. Sugumar, Greg Faanes et al. · Conference on High Performance Computing (Supercomputing) · 1994
Traditional supercomputers use a flat multi-bank SRAM memory organization to supply high bandwidth at low latency. Most other computers use a hierarchical organization with a small SRAM cache and a slower, cheaper DRAM for the main memory. Such systems rely heavily on data locality for achieving optimum performance. This paper evaluates cache-based memory systems for vector supercomputers. We develop a simulation model for a cache-based version of the Cray Research C90 and use the NAS parallel benchmarks to provide a large-scale workload. We show that while caches reduce memory traffic and improve the performance of plain DRAM memory, they still lag behind cacheless SRAM. We identify the performance bottlenecks in DRAM-based memory systems and quantify their contribution to program performance degradation. We find the data fetch strategy to be a significant parameter affecting performance, we evaluate the performance of several fetch policies, and we show that small fetch sizes improve performance by maximizing the use of available memory bandwidth. >
Cited by 27 works. doi:10.5555/602770.602815
Authors
References
- Sorting and Searching
- Cache Memories
- The NAS Parallel Benchmarks
- Structural aspects of the System/360 Model 85, II: The cache
- Using cache memory to reduce processor-memory traffic
- Data prefetching in multiprocessor vector cache memories
- Vector processing on the Alliant FX/8 multiprocessor
- Vector system performance of the IBM 3090
- Cache performance of vector processors
- Performance of cached DRAM organizations in vector supercomputers