Path: chuka.playstation.co.uk!news
From: "John Blackburne" <johnb@hk.super.net>
Newsgroups: scee.yaroze.freetalk.english
Subject: Re: mathengine SDK
Date: Tue, 02 Mar 1999 20:03:38 +0000
Organization: PlayStation Net Yaroze (SCEE)
Lines: 35
Message-ID: <7bhg6f$9he7@chuka.playstation.co.uk>
References: <7b9luf$9po25@chuka.playstation.co.uk>
NNTP-Posting-Host: th-pm01-25.ndirect.co.uk
Mime-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Newsreader: Microsoft Outlook Express Macintosh Edition - 4.5 (0410)

>  Anyone checked out the Mathengine SDK at www.mathengine.com  ?
> 
> Its a free to try game physics engine creator. Not sure if its possible to
> use it with yaroze though, only d/ling it now.

As it is it's not - it's x86 only. I can't imagine it or anything similar
making it onto the Yaroze (or PSX), as it doen't make sense to create a
generic physics engine for such hardware. To do so you would at minimum need
to use floating point representation, and without built in FP hardware the
PSX sucks worse even than a 80486 at FP.

The best solution on the Yaroze/PSX is an engine hard coded around the PSX's
hardware limitations to suit your particular requirements. This lets you
choose (e.g.) a fixed point representation for variables that makes sense in
your game universe, and lets you simplify/eliminate calculations and steps
where they will not be noticed in your engine, such as assuming all drag is
linear, or using constants to allow the compiler to pre-calculate expression
and sub-expression results.

I imagine such a generic engine would be very well suited to future consoles
such as the one announced by Sony today. which has horsepower enough to cope
with anything PCs can in FP, epecially vectored FP of the sort useful for
physics simulations. Such an engine might help developers working on
cross-platform titles, or who want to prototype on PC before porting to a
console.

But IMHO an even better solution is to hire a maths graduate with C
experience and set them to it, as the maths of such engines is hardly rocket
science. The one I'm working on already does rigid body simulation a lot
more sophisticated than the one described on mathengine's pages, which seems
to involve some pretty gross simulations such as all collisions are between
objects which can be treated as spheres. Fine for bouncing balls but not
much else.

John