The implementation of edge detection on HSA environment

Abstract

This paper presents the implementation of image edge detection on Heterogeneous System Architecture (HSA). HSA which includes ARM processor, Coprocessor and FPGA are compared with x64 CPU in terms of performance and power consumption. The experimental results show that although the best execution time is from x64 CPU, HSA has 50 times more energy efficiency. Also, HSA can exploit coprocessors and reconfigurable hardware to reduce processing time and it can achieve 1.3x speedup when process an image with the size of 512×512 pixels.