3D FPGA versus multiple FPGA system : enhanced parallelism in smaller area

Abstract

Handling large amounts of data is being limited by bandwidth constraint between processors components and their memory counterparts. Three-dimensional integration (3D) is providing possible solution to handle such critical applications. Especially for running larger designs when implementing on multi-FPGA platform, which can produce huge amount of fine-grain parallelism, for satisfying the speed and reliability needs 3D FPGA can be considered as a close candidate to choose. In this paper we tried to show the benefits of running larger designs on 3D FPGA compared to running on multi-FPGA systems using benchmark simulation. Results showed that a TSV-based 3D FPGA achieved better performance and area results.