Implementation of Distributed Processing Using a PC-FPGA Hybrid System

Recent research has focused on decreasing processing times and power consumption by combined use PCs and FPGAs. Here describe a system that performs distributed processing across multiple PCs and FPGAs using ethernet for PC connections and an in-house designed FPGA network to connect FPGAs. For processing software, we used Apache Spark on the PCs and an in-house application programming interface (API) for the FPGAs. Experimental results show that both PC and FPGA are effectively used for image compression calculation.