Along with Thu Nguyen, a graduate student, we are investigating how to support distributed real-time rendering -- the display of computer generated scenes in real-time using multiple nodes connected over a high speed network. The focus of this work is on resource allocation. We are currently concentrating on CPU, hardware graphics accelerator, main memory, and network.

Our goal is not to obtain the highest rendering performance possible on the most exotic equipment, but rather to improve performance on "commodity systems" by making use of network available resources. While our experimental work is currently targeted to a set of SGI O2's connected by 100mbps Ethernet, we plan to reimplement on PCs, assuming that we overcome a number of hardware and software problems on that platform that caused us to select the O2's in the first place.