[Haskell-cafe] Parallel graphics
noteed at gmail.com
Tue Sep 15 08:04:35 EDT 2009
2009/9/15 Andrew Coppin <andrewcoppin at btinternet.com>:
> I'm presuming that sparking every individual point is going to create
> billions of absolutely tiny sparks, which probably won't give great
> performance. We could spark every line rather than every point?
You should just try: no one can say if pixels, lines, image buckets or
complete images is the right granularity.
As for progress information, probably the needed granularity is not
the same as the above one.
Also, parallelizing a raytracer or the computation of a simple
function of coordinates won't need the same solution. For instance ray
coherence can be used to more efficiently traverse the accelarating
data structure of the raytracer with many rays at once.
Maybe before looking at what haskell has to offer you should look how
your program can be parallelized then see if your solution maps well
to one of haskell technical solutions.
More information about the Haskell-Cafe