On the Road Towards Superconductor Computers: Twenty Years Later

Abstract : Attempts to build superconductor computers based on low-temperature super-conductivity of Josephson junctions have come and gone during the last thirty years. In spite of both industry and the federal government's deep involvement in these efforts, no full-fledged superconductor computers have been built so far. As a result, superconductors, once considered an alternative to semiconductors, have lost their appeal to the general design community. By 2003, tremendous improvements in silicon-based technology have allowed semiconductor chips to have hundreds of millions of transistors and clock frequencies exceeding 3 GHz, speeds earlier considered the province of exotic technologies, such as super-conductors. Although superconductors have not become a mainstream digital technology (and perhaps it is fair to say they were never expected to be), they are still trying to find their way into several important niche areas, one of which is extreme (at the moment called petaflops) computing. Thanks to the support by various U.S. government agencies, work on superconductor technology and processor design continued behind the scenes, nearly invisible to silicon-based computer designers. In this paper, we consider the lessons learned from past superconductor computer projects, the current status, and future directions of the work in this field.