Ray Tracing Acceleration Through Heterogeneous Integrated Commodity Hardware

The ray tracing algorithm is a computationally intensive workload that has active efforts to improve its performance in the CPU and GPU architectures. In this research, we designed a specific implementation of the ray tracing algorithm for the APU architecture that has an integrated GPU and CPU in the same die. The APUs are regarded as commodity hardware. Our design uses the full internal resources of the APU to improve performance. We tested the classic and the BVH accelerated ray tracing algorithms and found that our implementation for the APU architecture bests the CPU and GPU solutions from the same price range in both cases.