High-end microprocessors are the epitome of integrated circuit technology. But it is low end microprocessors which fuel progress in microprocessors. More than 98% of the four billion microprocessors forecast to ship worldwide in 1995 will end up in low end embedded applications. The microprocessor industry spends most of its research dollars developing high-end microprocessors for CPU applications. Current high-end microprocessor implementations exploit most of the inherent parallelism in their instruction streams. Where does microprocessor development go from here? I give an overview of the development of the microprocessor industry and I show that to predict where the microprocessor is going one must first understand the industry's illogical behavior.

Summary form only given, as follows. Workstations dominate the high end of the desktop market. PCs own the rest of the desktop market. The desktop wars are being fought at the interface. Will workstations eat into the market for PCs? The technical issue is the struggle between old (established) architectures and new architectures. The author outlines the advantages and disadvantages of the old and new architectures and attempts to predict the outcome of the struggle for market share. Prediction is difficult because the deciding issues aren't technical: the computer industry is swept by fads just as the fashion and toy industries are. Fads and the trade press contribute to differences between folklore and reality. Folklore defines the 'interesting' problems and contributes significantly to the outcome of an issue. The author attempts to identify differences between folklore and reality and predict the outcome of the desktop wars.>