Re: Ray tracing in OCaml (revisited)

Jon Harrop <usenet@[EMAIL PROTECTED]
> writes:
> The OCaml implementation is now outperforming the C++ implementation
> compiled with g++ 4.2.3:
>
>   http://www.ffconsultancy.com/languages/ray_tracer/results.html
>
> Albeit, because the g++ compiled code is getting slower with newer
versions
> of g++...

Are these implementations close enough to even compare meaningfully?

[It seems like about 95% of the cpu time in raytracing goes to searching
for ray-surface intersections -- acceleration structure searching plus
primitive intersection tests -- so I suppose one could easily make a
fairly small test case that tested only those things, and make sure to
use the same algorithms for both languages, etc.]

-Miles

-- 
Everywhere is walking distance if you have the time.  -- Steven Wright