Re: [AGG] windows mobile

[Stefano Z. <ste...@ya...>]

No, I do not have any AGG_SBGR24, so my version is 2.5 or 2.4 (I was 
sure to has downloaded last relase from sourgeforce...strange ).

Anyway, it looks like AGG is faster but not so fast for an ARM 600Mhz an 
the fault is in my memory.

Effectively there isn't a unit measure for that. May the render of lion 
be the base unit but we do not have a base estimate with which compare.

Il 30/01/2013 00:51, Jim Barry ha scritto:
> On 29 January 2013 21:44, Stefano Zaglio <ste...@ya... 
> <mailto:ste...@ya...>> wrote:
>
>     I'm compiling AGG under windows mobile as I done in 2005.
>     I don't know if my memory is bad or if something is changed in the
>     source since then but the performance on a WM6 (ipaq with a
>     600Mhz) are
>     very poor.
>
>
> It's probably my fault. I think the most likely explanation is that 
> the examples have all been changed so that they render into a linear 
> RGB buffer, which is converted to an intermediate sRGB buffer before 
> being sent to the screen. The conversion to sRGB could potentially be 
> quite time-consuming on a low-performance device.
>
> You can get the old (incorrect, but faster) behaviour by changing the 
> "#define AGG_BGR24" at line 17 of trans_curve1.cpp to "#define 
> AGG_SBGR24". Let me know how much of a difference it makes.
>
> Cheers,
>
> - Jim
>
>