It never was the state of the art(or for a short time until Fermi came out). Even they themselves didn't use the 'persistent thread' thing anymore in their later paper, it didn't make sense on Fermi. I never quite understood their paper, and I never bothered to try out their persistent threads. My simple path tracer that actually did some real shading had a similar speed as their persistent threads approach at the time so I never bothered.shiqiu1105 wrote:I don't understand. If Timo Aila's work is the state of the art of gpu ray tracing, then why don't dicuss that in the thesis, instead talk about path regeneration and stream path tracing??
Must read and other references.