Scene generation fidelity requirements are reaching the limits of traditional raster-based rendering methods. Thus, a need has arisen to perform a thorough investigation and design for innovative approaches that include methods like ray tracing which can scale as a rendering solution. Signature Research, Inc. details a solution which relies on the development of a benchmarking data set and benchmarking framework. This new capability allows one to quantitatively evaluate rendering system designs, gauge implementation decisions, and navigate advances in hardware and software technology. Our resulting proof-of-concept real-time rendering system design is based upon the foundation of the collection of empirical data across three aspects of the problem. 1. Evaluation of a representative sample set of rendering frameworks covering open- and closed-source software, raster- and ray trace-based rendering methods, and various hardware and software optimization approaches. 2. Benchmarking hardware changes such as networking architectures, processor designs, and cache layouts. 3. Benchmarking software configuration changes at the operating system and rendering algorithm levels. These empirical results will then feed into our rendering system design to produce a rendering solution that can meet the performance and fidelity requirements for existing and future use cases.
