Hardware and software cache prefetching techniques for MPEG benchmarks

With the popularity of multimedia acceleration instructions such as MMX, MP EG decompression is increasingly executed on general purpose processors ins tead of dedicated MPEG hardware. The gap between processor speed and memory access means that a significant amount of time is spent in the memory syst em. As processors get faster-both in terms of higher clock speeds and incre ased instruction level parallelism-the time spent in the memory system beco mes even more significant. Data prefetching is a well-known technique for improving cache performance. While several studies have examined prefetch strategies for scientific and commercial applications, this paper focuses on video applications. Data is presented for three types of hardware-prefetching schemes: the stream buff er, the stride prediction table (SPT), and the stream cache, as well as a n ew software-directed prefetching technique based on emulation of the hardwa re SPT. Up to 90% of the misses that would otherwise occur with no prefetch ing are eliminated. The stream cache can cut execution time by more than ha lf with the addition of a relatively small amount of additional hardware. S oftware prefetching achieves nearly equal performance with minimal addition al hardware. Techniques presented in this paper can be used to improve perf ormance in a general-purpose CPU or an embedded MPEG processor. Performance gains achieved for MPEG benchmarks apply equally effectively to similar mu ltimedia applications.