Parallelising Ray Tracing on a KSR1 shared memory system


Standard and accelerated ray tracing have been also parallelised on a Transputer network and on a KSR1 with the collaboration of CNC, University of Manchester. An evaluation of these different architectures has been done.

The diagrams below show an analysis of computation time measured on a KSR1 versus number of threads and image tiling (centre), and volume subdivision and image tiling (right), of three test images (left).

Back to Home page