So now that S@H is open source, is it possible that we can have optimized
clients now for specific hardware like using Altivec on G4 and G5 chips?
It should be possible to still get the same results even when optimizing the
code.

I haven't looked at the code too much but there must be some things that can
be done to optimize the code.

Also, now that GCC(3.3?) has auto-vectorization, does anyone think that
this'll speed up the processing when S@H is compiled with it turned on?

Devon.