New Orleans, La. -- The future of graphics-oriented workstations in a market increasingly dominated by accelerated PCs was debated in a panel entitled "Graphics PCs Will Put Workstation Graphics in the Smithsonian" before a standing-room-only audience at the SIGGRAPH 96 convention. The participants were Michael Cox of PC accelerator manufacturer S3 Inc, Michael Deering of Sun Microsystems, Jay Torborg of Microsoft and Kurt Akeley of Silicon Graphics Inc (SGI). Cox began the discussion by observing that the accelerated PC/workstation competition represents a Darwinian process wherein the increased volume of PC sales would inevitably be accompanied by increasing incentives and innovation for the PCs. Citing developments such as embedded DRAM and Intel's AGP 0.5 GB bus dedicated to graphics, he asserted that the process would move PCs further and further up the pyramid of computational power, increasing both their market share and their threat to the present high-end dominance of workstations. Deering countered Cox's claims by noting that PCs, whether accelerated or not, were designed specifically for entertainment purposes as opposed to serious professional applications requiring extensive, intricate and highly accurate visualizations that are workstations' raison d'etre. Deering saw this as a lethal flaw in Cox's arguments: the increasing volume of PC sales would thus encourage nothing more than the lowest common denominator of computational capability acceptable to the mass market. In keeping with this notion, he pointed out that impressive displays generated by PCs were simply products of large texture maps giving the illusion of detail. In addition, he castigated PC advocates for mendaciously comparing what workstations presently are with what accelerated PCs want to become. Torborg picked up and attempted to strengthen Cox's thread. In particular, he claimed that ruthless economics of scale would eventually win the day: the rising volume of PC sales would necessarily produce more firms engaged in PC- related technology and thus a burgeoning price/performance ratio for PCs. Noting that the current workstation/PC conflict is analogous to the former supercomputer/workstation competition, he predicted that, since similar market forces are operational in both cases, workstation manufacturers would go the way of former supercomputer-makers, as more participants in the PC market would pioneer innovation and an open-platform approach. Akeley, in turn, rounded out the discussion by bolstering Deering's position. Specifically, he observed that workstation graphics are simply and clearly superior to those of PCs -- and that today's high-end applications cry out for even more sophisticated graphics capabilities, not less. He strongly contested the notion that increasing volume drives innovation, asserting instead that PCs, as archetypal commodity products, represent the very antithesis of innovation. He said that workstations, as opposed to PCs, were value-added products, and that one could not do truly innovative work "on a 1-year product cycle". Akeley concluded by noting that: (1) although PCs are ubiquitous, workstations lead, because they represent the first and best models; (2) supercomputers are still sold; (3) the market pyramid is not a static entity but getting larger faster -- thus insuring a significant permanent role for workstations at the high-end, even as PCs move up. Judging by the vigor of audience response, sympathies lay decidedly with Deering and Akeley. Certain provocative issues, however, were not placed under intense scrutiny, e.g. the roles to be carved out by systems like DEC Alpha running NT or workstations that will be constructed from COTS components.
SIGGRAPH PANEL DEBATES FUTURE OF WORKSTATIONS
August 16, 1996