It seems like this entire pipeline could be turned into code (like we are doing with the renderer in assignment 2), instead of having certain portions implemented in hardware. Is this where GPUs are heading? Or would this sacrifice to much graphics performance?
This comment was marked helpful 0 times.
kayvonf
@mofarrel. A great question that researchers continue to wrestle with. The question is whether the loss of efficiency (performance, power, etc.) is justified by the greater flexibility afforded by implementing more of the pipeline in software. Some friends and I made one attempt, called GRAMPS a couple of years back. Some of my Ph.D. students are continuing to think about this question.
It seems like this entire pipeline could be turned into code (like we are doing with the renderer in assignment 2), instead of having certain portions implemented in hardware. Is this where GPUs are heading? Or would this sacrifice to much graphics performance?
This comment was marked helpful 0 times.
@mofarrel. A great question that researchers continue to wrestle with. The question is whether the loss of efficiency (performance, power, etc.) is justified by the greater flexibility afforded by implementing more of the pipeline in software. Some friends and I made one attempt, called GRAMPS a couple of years back. Some of my Ph.D. students are continuing to think about this question.
This comment was marked helpful 0 times.