7.2   THE PIONEERS

An idea for a FPGA as distinct from the programmable logic device (PLD) was developed by Ross Freeman, an engineer with Zilog, Inc., a small semiconductor-chip-making subsidiary of Exxon Corporation. Freeman’s idea was not warmly received by the executives of the parent company and as a result three individuals, Bernard Vonderschmitt, Ross Freeman, and James Barnett left Zilog, raised $4.25 million in venture capital from Hambrecht & Quist and Kleiner Perkins Caufield and Byers, among others, to form Xilinx, Inc., in 1984. The previous year, Altera Corporation was founded to provide fuse-programmable alternatives to mask-programmable gate arrays, and was first in the market. However, Xilinx’s patented static-random-access-memory-based (SRAM) FPGA architecture was a more radical development and soon outpaced the Altera product. At the same time, two additional companies were founded: Atmel Corporation (1984) and Actel Corporation (1985), both manufacturers of programmable semiconductor components. Atmel Corporation designs and manufactures PLDs and FPGAs, as well as mask-programmable gate arrays. Actel Corporation became the leader in the development of FPGAs based on antifuse switching elements. In 1994 Actel was the world’s leading antifuse FPGA producer and held number two market position in FPGAs, just behind Xilinx.

FPGAs were originally introduced to the marketplace by start-up companies rather than established semiconductor suppliers. This was consistent with the pattern for earlier components, such as memory, microprocessors, and programmable logic. Indeed the parallel with the microprocessor “revolution” in the late 1970s is particularly relevant. As a rule, large organizations have difficulty introducing new components. Often they are unwilling to take the speculative risks required even though new product development may enhance sales of existing components and systems. As with Xilinx and Actel, the role of venture capital has been critical, particularly for start-up companies in the United States, along with the commitment of individual engineers willing to risk career and family security, and devote the effort required to transform a laboratory brainchild into a successful commercial product.

Several factors enabled the FPGA to become a successful asset to system designers. From the outset, the FPGA had to be supported by proven CAD software, as well as provide competitive solutions to existing implementation problems. At the same time, the emergence of electronic CAD (ECAD) as a business in its own right, allowed the entrepreneurial FPGA companies to use existing design entry software and to concentrate on perfecting the hardware. The reshaping of the integrated-circuit (IC) industry into specialist operations, for example, memory companies and “Silicon foundries” providing reliable fabrication of FPGAs as well as application-specific integrated circuits (ASICs), also served to pave the way for the FPGA. Other factors that aided the success of FPGAs included:

• Widespread availability of powerful, yet inexpensive personal computers and engineering workstations.
• Development of electronic CAD for IC technologies and design styles, including placement, design-rule checking, partitioning, routing, and simulation.
• Development of systematic approaches to digital design, including finite state machines, datapath design, and timing disciplines.