An Area-Efficient FPGA Overlay using DSP Block based Time-multiplexed Functional Units (1606.06460v1)

Abstract: Coarse grained overlay architectures improve FPGA design productivity by providing fast compilation and software-like programmability. Throughput oriented spatially configurable overlays typically suffer from area overheads due to the requirement of one functional unit for each compute kernel operation. Hence, these overlays have often been of limited size, supporting only relatively small compute kernels while consuming considerable FPGA resources. This paper examines the possibility of sharing the functional units among kernel operations for reducing area overheads. We propose a linear interconnected array of time-multiplexed FUs as an overlay architecture with reduced instruction storage and interconnect resource requirements, which uses a fully-pipelined, architecture-aware FU design supporting a fast context switching time. The results presented show a reduction of up to 85% in FPGA resource requirements compared to existing throughput oriented overlay architectures, with an operating frequency which approaches the theoretical limit for the FPGA device.