3 ms·

This is one of my favorite graphics/rendering papers from 2017; great work from the entire DWA Moonray team! In addition to the main paper, there are some other

by ykl 9y ago

This is one of my favorite graphics/rendering papers from 2017; great work from the entire DWA Moonray team! In addition to the main paper, there are some other materials that were presented at HPG and SIGGRAPH 2017 by the same team. HPG Slides: http://www.highperformancegraphics.org/wp-content/uploads/2017/Papers-Session5/HPG2017_VectorizedProductionPathTracing.pdf http://www.highperformancegraphics.org/wp-content/uploads/20... SIGGRAPH Course Notes (the relevant part begins on Page 35): https://jo.dreggn.org/path-tracing-in-production/part1.pdf https://jo.dreggn.org/path-tracing-in-production/part1.pdf SIGGRAPH Slides: https://jo.dreggn.org/path-tracing-in-production/MoonrayV3.pdf https://jo.dreggn.org/path-tracing-in-production/MoonrayV3.p... Also, the actual shading system they used was also presented at SIGGRAPH 2017: http://blog.selfshadow.com/publications/s2017-shading-course/dreamworks/s2017_pbs_dreamworks_notes.pdf http://blog.selfshadow.com/publications/s2017-shading-course...

pixel_fcker 9y ago
It's interesting work but it does seem like the speedup they get is pretty poor (1.3x to 2.3x) when going from 1 to 16 SIMD lanes. It seems like the overhead from all the queue management and AOSOA transformation must negate most of the benefit of the parallelization? They also mention the fact that programming the system is hard, and plugins must fall back on single-lane code paths until they can be coded into the system proper. I would assume all the ray sorting makes it extremely difficult to use any bidirectional methods as well.