A Performance Comparison of Path Tracing on FPGA and GPU

Abstract: Ray Tracing algorithms have long been the popular choice for rendering realistic images, and in recent years they have also reached the field of real time computer graphics. Although their performance has seen great improvement, they are still very computationally costly to perform, both in terms of time and power. Improving their performance and efficiency is therefore a very relevant field of research, to improve current applications and open up new possibilities for the future. In this degree project we have focused on implementing a ray tracing algorithm called Path Tracing on an Field Programmable Gate Array (FPGA), a type of reconfigurable hardware. FPGAs provide a new path towards increasing performance as we approach the end of Moore’s Law and reach the limits of general-purpose computing, making them an interesting candidate for ray tracing. This study resulted in a substantial performance difference between FPGA and GPU, with GPU outperforming FPGA by around 20 times in full path tracing.