Re: [Devel] Optimizatins to ttinterp.c - floating point is bad on ARM

[Graham . Asher]

Floating point is *very* slow on ARM processors so should be avoided for
platforms like EPOC which predominantly run on them.

This may be true for other processors used in phones, PDAs and the like.

Graham Asher
Symbian Ltd.



                                                                                
                                             
                    Nathan Hurst                                                
                                             
                    <address@hidden        To:     David Williss 
<address@hidden>                          
                    sh.edu.au>                     cc:     address@hidden       
                                         
                    Sent by:                       Subject:     Re: [Devel] 
Optimizatins to ttinterp.c                       
                    address@hidden                                              
                                    
                    g                                                           
                                             
                                                                                
                                             
                                                                                
                                             
                    06/12/00 22:09                                              
                                             
                                                                                
                                             
                                                                                
                                             




On Wed, 6 Dec 2000, David Williss wrote:

> This is done via the __inline keyword, which I would create a #define
> for if I cared about portability. While I was at it, I added __inline
> to a few other really short functions but they didn't have much
> impact.

Is there 'inline' keyword support in your compiler?  that is the standard
way to indicate inlining.

#define inline __inline

> The big gain came from TT_MulDiv().  It turns out that all the other time
> soaking functions end up calling it.   Since TT_MulDiv is exported from
> the library, this would need some cleaning up, but it boils down to...

I'm wondering whether it would be better to use a floating point
version for most high end machines?  (once we have a glyph, the render
would still be better in fixed point, but there is a lot of chance of
bugs in complex code that performs scaling or sqrts using fixed point
operations for things like glyph extent calculation.)

njh



_______________________________________________
Devel mailing list
address@hidden
http://www.freetype.org/mailman/listinfo/devel