[ODE] Sweep-and-Prune space for ODE

[Graham Fyffe]

Okay, now THAT is a nice implementation of radix sort.  For floats, it
is 2.3 times faster than std::sort!  That means it's 5 times faster
than Pierre's code.  Five.  I would strongly suggest that anyone using
Pierre's code should switch to the stereopsis code.

- Graham


On Fri, 26 Nov 2004 00:21:56 -0000, Simon Barratt
<barog at pineapple-interactive.com> wrote:
> I'm not sure about OPCODE's radix sort speed wise but this link might be
> interesting..
> 
> http://www.stereopsis.com/
> 
> It contains a "Radix sort for floating point" page which has some
> optimisations based on OPCODE's radix sort and claims to get some good
> speed boosts, it may be doing some floating point tricks which might
> need #ifdef'ing to not scare some people away.
> 
> Hope that helps!
> 
> --
> Simon Barratt, Lead Developer, Pineapple Interactive Ltd
> e: barog at pineapple-interactive.com
> w: www.pineapple-interactive.com
> 
> 
> 
> -----Original Message-----
> From: ode-bounces at q12.org [mailto:ode-bounces at q12.org] On Behalf Of
> Graham Fyffe
> Sent: Thursday, November 25, 2004 5:13 PM
> To: Brian Marshall
> Cc: ode
> Subject: Re: [ODE] Sweep-and-Prune space for ODE
> 
> On the subject, I did try several implementations of Radix Sort, and the
> fastest one for integers was twice as fast as std::sort.  If I took that
> implementation, and added Pierre's trick for handling signed floats,
> maybe it would still beat std::sort after all...
> 
> - Graham
> _______________________________________________
> 
> 
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
> 
>