Ray Tracing on Graphics Hardware

Ray tracing is one of the important elements in photo-realistic image synthesis. Since ray tracing is computationally expensive, a large body of research has been devoted to improve the performance of ray tracing. One of the recent developments on efficient ray tracing is the implementation on graphics hardware. Similar to general purpose CPUs, recent graphics hardware can perform various kinds of computations other than graphics computations. One notable difference between CPUs and graphics hardware is that graphics hardware is designed to exploit a larger degree of parallelism than CPUs. For example, graphics hardware usually processes several thousands of independent pixels (compare it to a few independent tasks on multi-core CPUs). Therefore, the key is to use algorithms that are suitable for parallelization in order to harness the power of graphics hardware. Although the parallel nature of ray tracing seems to be well suited for graphics hardware, there are several issues that need to be solved in order to implement efficient ray tracing on graphics hardware. This paper surveys several algorithms for ray tracing on graphics hardware. First, we provide an overview of ray tracing and graphics hardware. We then classify several algorithms based on their approaches. We also analyze the bottlenecks of current approaches and present possible solutions. Finally, we discuss future work to conclude the survey.