The research paper on GPU Ray Tracing (http://dl.acm.org/citation.cfm?id=2447976.2447997&coll=portal&dl=ACM) coauthored by Prof. Morgan McGuire is featured in the May 2013 Communications of the ACM. This is part of a regular series in which every month the ACM highlights one or two papers selected from across all of computer science that are considered groundbreaking work for the field.
This paper describes the key algorithms behind the NVIDIA OptiX product, which is a system for tracing geometric rays through space using massive parallelism on graphics processors. It has been applied to problems as diverse as real-time generation of photorealistic images, audio simulation, and collision detection. OptiX demonstrates that it is possible to create high-performance systems that are themselves programmable through the use of JIT compilation and eliminating the user program / system software boundary.
In his discussion of this work (http://dl.acm.org/citation.cfm?id=2447996&bnc=1), Matt Pharr of Google says, “This paper is a must-read for anyone who cares about writing extensible software systems that are also high-performance software systems. Although the target hardware architecture for this work is GPUs, the underlying ideas are equally applicable to high-performance software systems on CPUs.”
Megamenu Social