new radeon tiling patch

[Roland Scheidegger]

Alex Deucher wrote:
> I'll give this a test on mergedfb in the next day or so.  I quickly
> perused the DDX changes and I noticed this:
> 
> +    if (info->allowColorTiling && (pScrn->virtualY > 2048)) {
> +	xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
> +		   "Color tiling not supported with virtual x resolutions larger
> than 2048\n");
> +	info->allowColorTiling = FALSE;
> +    }
> 
> you'll probably want to check pScrn->virtualX as well.  maybe something like:
Good catch. virtualY is a typo actually, only virtualX needs to be below 
2048. It _should_ (untested) work with virtualY larger than 2048. (I've 
replaced the ddx8 patch with the fixed version.)

> Looks good so far otherwise.
I forgot to mention in the last message, I've also done some quick tests 
on a radeon 7200 sdr. Doesn't look that good. Well, everything runs 
correct, but the performance boost resulting from color tiling is very 
minimal. In contrast to the 30-40% on the radeon 9000pro I got in q3a, 
the 7200sdr only gained around 5% (I've tested other apps, and that 5% 
was pretty consistent). I'm not sure why exactly it doesn't benefit 
much, maybe color tiling was more an afterthought and the chip also 
tuned to work reasonable without it, whereas the rv250 might have its 
caches/memory accesses more tuned for color tiling.

Roland