Applications

In 1980 there weren’t many parallel applications. Those that existed were mostly experimental in research labs. But then, parallel machines were few and they cost millions of dollars. Now there are many parallel applications and we have to select a small representative set to cover in this chapter, see Table 1.1 or 4.1. By 1980 most of these applications had already made a start in parallel processing but the progress was very different. Parallelism in seismic processing started in the 1960s on IBM attached vector processors and was using vector supercomputers on production seismic surveys by 1980. Finite element analysis and computational chemistry were beginning to see their potential [GlHo83, FiRa81]. As early as the 1960s computer vision researchers saw that 2D processor arrays such as Illiac III could work well for image processing [McCo63]. Object recognition, on the other hand, had not seen much opportunity for parallelism because algorithms were still being explored. In 1980, computer graphics experts were experimenting with graphics systems on the Cray-1 and on a DMM system [EwMa78, Fuch77]. By 1981, the E&S CT5 flight simulator came out which had a pipeline of custom VLIW processors controlled by a PDP-11. The whole simulator cost around $20M when a Cray-1 cost $8M. Bioinformatics was researching laboratory equipment and algorithms to support them, a long way from having any interesting results.