From Factory to Farm: Dissemination of Computing in High-Energy Physics

Abstract

Computing in high-energy physics relied on large computers and dedicated central facilities in the 1960s and 1970s. With the advent of colliding-beam accelerators, minicomputers, and microcomputers in the 1980s, physicists elaborated a new decentralized model of computing in high-energy physics. They restored control to and autonomy of their analysis of physics data, while simultaneously replacing large computers with farms, or clusters of microprocessors custom-built for their tasks. This new division of labor has become dominant in scientific collaborations in other fields and is presented here in the context of the culture of the Fermi National Accelerator Laboratory, which was a crucible for the creative efforts required to disseminate computing power to users. Fermilab, the last National Accelerator Laboratory, came on line as federal expenditures for high-energy physics ebbed in the 1970s and 1980s.