HIGH-PERFORMANCE IMAGE COMPUTING WITH MODERN MICROPROCESSORS

Various levels of parallelism have recently been introduced in advance d microprocessors to meet the demanding computing need in digital vide o processing and other multimedia applications. Because many imaging a lgorithms are easily parallelizable, these architectural features and their wide availability at low cost have become a powerful tool in tac kling both existing and new imaging applications. At the lowest level, the subword parallelism is used in the new instructions aimed at proc essing multiple multimedia data simultaneously. Instruction-level para llelism including subword parallelism is realized in either very long instruction word or superscalar architectures, while on-chip and/or of f-chip multiprocessing capability is available for easier multiprocess or system designs. One of the difficulties in maximizing the computing throughput via parallelism has been the level of programming in that to obtain the optimal performance, assembly-level programming has typi cally been required. We review the architectural features in several m odern microprocessors such as TMS320C60, TM-1000, PowerPC 604, Pentium Il, R10000, Alpha 21264, PA-RISC 8200, UltraSPARC-II, and TMS320C80. Various obstacles to obtaining the best performance from these micropr ocessors with high-level and assembly languages are discussed, and sev eral approaches to overcome these difficulties in diverse imaging appl ications are presented. (C) 1998 John Wiley & Sons, Inc.