Warp wrote:

>   The speed of Crays don't come from CPU speed, but from havint a *lot*
> of them (as well as fast memory buses, etc).
> 
>   POV-Ray versions previous to 3.7 do not benefit from a Cray at all.
> You get better speed by buying an old PC for $50.
> 
>   Now, POV-Ray 3.7 is a completely different story.

According to Wikipedia, the Cray-1 does an average of about 136 MFLOPS, 
but can peak up to 250 MFLOPS if the code is structured right.

I think in general most "supercomputers" only achieve peak performance 
for very specific kinds of workload. Just grabbing the POV-Ray source 
code and throwing it at a C compiler is *highly* unlikely to just happen 
to produce the right kind of workload.

(As you say, both Cray-1 and Cray-2 work because they're very parallel - 
they're vector-processing machines. POV-Ray isn't structured that way. 
It could be, but it isn't at present.)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message