ra2@doc.ic.ac.uk (Roger Attrill) (12/24/90)
The fact is it's not yet fit for public release anyway, Besides it's not fast
enough for me yet, and I want it *really* fast.
It's written for the Acorn Archimedes computer using the increasingly
popular ARM RISC Assembler language.
My degree takes precedance over finishing it at the moment, so I don't see
it being released in the near future. Maybe I'll get some time over the
summer. I only started it because I wanted to learn about ray-tracing, and
It was basically a programming exercise, to see how fast the Archimedes
*really* was. It's fast! The 'exercise' has grown out of all proportions
though, I never intended to write the whole thing in Assembler. The compiled
code is partially optimised at the moment, and is only about 40K. No
Libraries are used - I wrote *Everything* from scratch, My own versions of
m/calloc, floating point routines, transcendental functions, the lot! and
it's all included in the 40K code.
Incidentally, while I'm about it, I enquired to this group about the Schell
metzner sort routine for sorting surfaces, about this time last year. I've
recently written a *very* optimised quicksort routine in the Assembler
language mentione above. It's faster than the S-M sort by about 60% or so.
My routine sorts 100,000 integers in about 1.6 seconds on a 12mip (30 MHz)
Archimedes! (with ARM 3). I'll post if the interest is there.
Merry Chistmas,
Roger.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
| Roger C. Attrill | I don't always think therefore I'm not necessarily. |
| ra2@doc.ic.ac.uk | |
| Imperial College | and other variations on a theme. |
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-