Stardent and the Elusive Graphics Supercomputer 331
Table 12-2 illustrates the multiplicity-of-companies principle. Examples include Univac (now Unisys since it merged with Burroughs) and IBM in the early 1950s; TI and Fairchild, which started the semiconductor industry; DEC and SDS (now defunct) for minis; Cray and CDC for supercomputers; Apple, Altair and Commodore; Apollo and Sun for workstations; Alliant and Convex for minisupercomputers; and Ardent and Stellar, which created graphics supercomputers.
SILICON GRAPHICS
Silicon Graphics started up in the early 1980s to provide 3-D graphics terminals that were connected to minis and mainframes. By 1985, when more powerful 68000 microprocessors became available, they were expanded into 3-D workstations. The basic design for the Geometry Engine chip, which transformed polygons into a 3-D space, was completed in 1982 by SGI founder Jim Clark while at Stanford. Clark, an alumnus of the University of Utah, acquired the basic design ideas from his own thesis work and the powerful, highly pipelined Evans and Sutherland displays. In 1986, SGI adopted the MIPS architecture for its computational engine and began supplying a range of products from low-cost, diskless 3-D workstations to multi-processor workstations and computational servers.
STARDENT AND THE ELUSIVE GRAPHICS SUPERCOMPUTER
In January 1986, Stellar and Ardent (originally called Dana) started up in Belmont, Massachusetts, and Sunnyvale, California, respectively. The basic plan of both companies was to create a new computer class that would have substantially higher computing and graphics performance than existing workstations.
Ardent started with a $12.5 million first-round investment based on a plan put together by its seven founders during a month-and-a-half-long, accelerated, self-funded seed stage. Stellar obtained roughly the same level of funding based on a plan that had been in gestation for roughly a year.
Ardent's ten-page business plan, dated January 20, 1986, identified a number of risks that might jeopardize the plan. Table 12-3 shows the assumptions Ardent made in its plan regarding each of these potentially problematic areas and contrasts those assumptions with the actual outcomes.
In 1987, Ardent secured the second round of funding from Kubota Ltd., which took on the following functions: manufacturing the products in Japan, marketing in