Re: [hatari-devel] Is there a need to keep 2 nearly identical functions into hatari ?

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

Thanks for this reply:


> It's OK to me to have just one (some numbers of the slowdown
would be nice though).

If nobody is against, I'll remove the rendering_no_zoom one this week.
Thomas, do you agree too (I think you're the one who implemented the 2 functions).


>  There are some optimizations that could be done to the current code:

* Make VIDEL_memset_* functions "static inline" instead
  of just "static".

Ok, I'll do it


* use a variable instead of regetting bgcolor inside loop:
	Uint8/16/32 bgcolor	HostScreen_getPaletteColor(0);

  (compiler cannot know that it's always the same value.)

I agree here, but later, I'd like to implement the "rasters" (changing of the color of the border0).
I will read the value from a table.
So, I've not created the variable actualy to remove it later.


Laurent


 - Eero





Le 26/02/2012 21:19, Eero Tamminen a écrit :
> Hi,
>
> On perjantai 24 helmikuu 2012, Laurent Sallafranque wrote:
> > I think there's no really need to keep both render_screen_noZoom and
> > render_screen_Zoom functions.
> >
> > They both do the same thing, code is nearly identical.
> >
> > I can call all the time render_screen_zoom which works in all case.
> > It will slow a little when both zoomX and zoomY = 1, but half the time,
> > hatari runs with the zoom renderer (every time I'm in 320*xxx for
> > example, hatari uses the zoom renderer).
> >
> > I think it would be easier to keep only one.
> It's OK to me to have just one (some numbers of the slowdown
> would be nice though).
>
>
> There are some optimizations that could be done to the current code:
>
> * Make VIDEL_memset_* functions "static inline" instead
>    of just "static".
>
> * use a variable instead of regetting bgcolor inside loop:
> 	Uint8/16/32 bgcolor	HostScreen_getPaletteColor(0);
>
>    (compiler cannot know that it's always the same value.)
>
>
> 	- Eero