Power analysis of sorting algorithms on FPGA using OpenCL

With the advent of big data and cloud computing, there is tremendous interest in optimised algorithms and architectures for sorting either using software or hardware. Field Programmable Gate Arrays(FPGAs) are being increasingly used in high end data servers providing a bridge between the flexibility of software and performance benefits of hardware. In this paper we look at implementations of some of the most popular sorting algorithms using OpenCL which take advantage of FPGA architecture. We evaluate these implementations in terms of power consumption which is measured using dedicated server power loggers and execution on Intel Arria 10 hardware. Our experiments show that taking advantage of software FIFOs have a significant impact on power consumption as well as requiring less hardware and memory resources.