[ODE] ANN : Update ODE distribution!!!

[Dino Patti]

I saw this clean way of doing it, the other day ;)

http://www.flipcode.com/cgi-bin/msg.cgi?showThread=Tip-AutomaticCPUDetection&forum=totd&id=-1

It's really simply, and you only have to run the SSE detecion one time ;)

/*Dino*/



----- Original Message ----- 
From: "Adam D. Moss" <adam at gimp.org>
To: "ode" <ODE at q12.org>
Sent: Friday, October 17, 2003 3:04 PM
Subject: Re: [ODE] ANN : Update ODE distribution!!!


>
>
> Nguyen Binh wrote:
> >      Russ! What's your ideas on an SSE code part for ODE? My 2 cents
> >      is use another #define ODE_USE_SSE in config. Is it OK?
>
> I know you're not asking me, but personally I like any code
> that makes ODE faster.  However, without some sort of runtime
> architecture auto-detection (or at least a switch) it's of
> limited practical use for *me* -- for example, I don't want my compiled
> app to only run correctly on SSE-using processors, but it'd
> be nice if ODE could use SSE on processors where SSE is available.
> The idea of distributing multiple copies of ODE (with / without
> SSE/3dnow/whatever) with an app and run-time linking ODE dynamically
> doesn't particularly appeal, but might be doable (but isn't at all
> attractive while the performance benefits aren't large).
>
> If I could choose between algorithmic improvements (ie. stepfast) and
> implementation improvements (ie. SIMD-ing) I'd usually prefer the
> algorithmic improvements, being of universal appeal -- but SIMD-ing
> of ODE seems inevitable at some point. :)
>
> Regards,
> --Adam
> -- 
> Adam D. Moss   . ,,^^   adam at gimp.org   http://www.foxbox.org/   co:3
> "i18n ought to be abbreviated i2n to make it quicker to write!"-snout
>
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>