Re: A86: Line routine


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

Re: A86: Line routine




At 21:53 1998-05-05 -0400, you wrote:
>
>On Wed, 6 May 1998, Jimmy Mårdell wrote:
>
>> How fast? I've made a pretty fast line routine using self-modifying
>> code, no temporary mem and only one push/pop in the repeat routine.
>> It's about 4 times faster than Stephane Jantzens line routine.
>
>The inner loop of my line routine takes about 110 t-states per pixel (more
>or less, depending on the slope of the line).  It also uses no temporary
>memory and only one push/pop.  Do you want to exchange routines?  They
>probably look the same. :-) 

I counted the T cycles in the FLINE inner loop:

 * delta_x >= delta_y   ==>   98.5   T cycles average
 * delta_x <  delta_y   ==>  106.75  T cycles average

I think FLINE is available at ticalc.org (at least as an Usgard function
library).

--
Real name: Jimmy Mårdell                "can't go mucking with a 'void *'"
IRC......: Yarin                                // Apple's MPW C compiler
Email....: mailto:yarin@acc.umu.se        
Homepage.: http://www.algonet.se/~mja/


References: