Processor speeds are increasing much faster than memory speeds, and thus memory bandwidth is rapidly becoming the limiting performance factor for many applications, particularly those whose inner loops linearly traverse streams of vector-like data. Because they execute sustained accesses, these streaming computations are limited more by bandwidth than by latency. Examples of these kinds of programs include vector (scientific) computations, multi-media compression and decompression, encryption, signal processing, image processing, text searching, some database queries, some graphics applications, and DNA sequence matching.