Interconnect technology dictates architecture choices in FPGAs

Today there exist more than half a dozen different FPGA implementations with distinctly different architectures. The FPGA market has been characterized by proprietary tools and architecture-specific design practices that users must learn to take advantage of the resources on each architecture. FPGA designers have come to accept that they have to master the architectural details in order to achieve design objectives. In this paper, we discuss the evolution of FPGA architectures from pass-transistor interconnect based (1000-5000 Ω) FPGAs to two generations of antifuse-interconnect based (500-1000 Ω→50-100 Ω) FPGAs and how these choices impact the way we design