| Re: [AD] Poor Ming performance |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
> How can this be? If the hline() h/w is faster than hline() software,
> then wouldn't the primitive which depend on it also follow the same
> speed pattern? Wild guess here: is the bitmap locked/unlocked only once
> per primitive?
Of course !
The bitmap is locked for each software-rendered hline in the hline() test,
whereas it is locked for each filled circle in the circlefill() test. Once
the surface is locked, it seems that drawing directly onto it is much faster
than asking the gfx processor to do it for us.
--
Eric Botcazou
ebotcazou@xxxxxxxxxx