|
 |
On Wed, 06 Oct 2004 18:01:48 -0400, Noam Lenz wrote:
> A new algorithm has been developed at UCSD which accelerates certain
> graphics computations by two or three orders of magnitude and is simple,
> requiring only a few extra lines of code.
> The entire story can be found here:
> http://www.physorg.com/news1440.html
> I hope this can be used to accelerate povray.
Although there are many things that could be done to accelerate POV-Ray, I
do not think that this is one of them. This optimization only works when
the algorithm you are using already takes advantage of data parallelism,
like the GPU does in shading fragments. POV-Ray, overall, does not
resemble this type of algorithm at all, so the optimization does not
apply.
An more useful avenue of research might be to see how we can make more use
of data parallelism in POV-Ray, for example by tracing rays in bundles:
http://www.eg.org/EG/CGF/volume20/issue3/paper173
Andrew
Post a reply to this message
|
|
