An Optimization Method of WebP Images Lossy Compression Algorithm for FPGAs

To improve the efficiency of WebP images lossy compression algorithm for heterogeneous accelerating based on FPGAs in data center server, this paper proposed an optimization method of WebP image lossy compression algorithm for FPGAs. By adopting the optimization method of this paper, the overall throughput performance of WebP heterogeneous acceleration solution based on OpenCL language can be up to 281MPixel/s. The energy efficiency ratio is 2.29x higher than Intel Xeon E5-2690v3 processor. Compare with the industry-leading WebP heterogeneous acceleration solution of Xilinx and CTAccel, the throughput performance of this paper is 2.5x better than Xilinx and 1.8x better than CTAccel. Finally, this paper improves the utilization of RAM resources according to the characteristics of the RAM block in the FPGA.