chicken-hackers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: regarding floating-point performance - Re: [Chicken-hackers] 3D engi


From: Brandon J. Van Every
Subject: Re: regarding floating-point performance - Re: [Chicken-hackers] 3D engine strategy
Date: Fri, 23 Feb 2007 11:12:41 -0800
User-agent: Thunderbird 1.5.0.9 (Windows/20061207)

Tony Sidaway wrote:
On 2/23/07, Brandon J. Van Every <address@hidden> wrote:
Tony Sidaway wrote:
>
> I don't really see that there's a problem here.  What's wrong with
> implementing a fp array type and writing the heavy matrix arithmetic
> in an optimized C module?

Could you provide some more context for your thinking here?  It's not
clear to me what kind of computational problem you're talking about.
Thus I can't comment on whether partitioning the problem between Scheme
and C is easy or onerous.

> C takes care of all the low-level crap and
> the beauty of Chicken is that you still have Scheme for the high level
> stuff.
>

In my opinion, computational geometry is both high and low level stuff,
often in the same context.  One shouldn't be burdened with
re-implementing the code in C.

I think it depends on your desired outcome.   I'm guessing that if
you're concerned with fp performance and optimization in particular,
you must want to do large repetitive sequences of floating point
calculations.  The logical way to handle this in Chicken is to put the
sequences into vector or matrix form and use a suitable C or C++ math
package to perform the transformations you want.  But I know nothing
about modern game production so I don't know whether this would answer
your concerns.  It would probably be very easy to do, though.

My experience causes me to disagree with you. Unless you want to refer to a more specific 3D problem, we'll just agree to disagree. C is not a flexible language, and many computational geometry problems need both flexibility and performance at the same time.


Cheers,
Brandon Van Every





reply via email to

[Prev in Thread] Current Thread [Next in Thread]