Fwd: Re: [Shootout-list] Science-related benchmarks

[Jon Harrop]

On Wednesday 27 April 2005 10:03, Sebastien Loisel wrote:
> > > something multi-dimensional involving either LU, sparse LU or GMRES
> > > and other algorithms.
> >
> > Excellent. Where can I find more details on these?
>
> ...
> So I'd like to write either an implicit PDE solver or a simplified
> rigid body simulator.

Thanks for the links. What code have you written so far?

> > In addition to polymorphism, I'd like to see implementations of solutions
> > which naturally use higher-order functions (e.g. Newton-Raphson). This is
>
> My intended tests (including implicitode) use that kind of stuff.

That's covered then. :-)

> > > I don't think they'd be faster, but the algorithm would be more
> > > representative of a real algorithm.
> >
> > We'll have to disagree here. In my work, my optimisations are almost
> > always
>
> Hey? I was saying it was unlikely for people to render the snowball
> flake, and more likely for them to render some grassy knoll, for which
> you'd use hierarchical bounding volumes.

The optimisation I proposed is general purpose, not specific to the sphere
flake. I've written the ray tracer now (currently 224 LOC in OCaml) but used
hierarchical spherical bounding volumes instead of the set-plane trick as
these can be used to accelerate non-primary rays. It can generate and ray
trace a sphere flake containing over half a million objects with shadows and
reflections in 36s at 1280x768.

Code for the shootout could be a lot shorter as my current implementation
does incremental rendering using OpenGL and the input is generated in the
same program. I'm not sure I can get it below 100 LOC though.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
Objective CAML for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists