2010/12/14 Aaron J. Seigo <ase...@kde.org>: > On Tuesday, December 14, 2010, Benoit Jacob wrote: >> Different users have different priorities. For example, Gecko wants >> pixel-exact rendering > > risking being "pedantic" ... this isn't actually true. it's true most of the > time, but some of the time pixel perfect is irrelevant. "*gasp*, what?" > > case in point: bluring. > > when a blur is applied, it's not important to ensure "pixel perfection" for > what is being blurred.
Sure, I never said that everything needed to be pixel-exact. And many things only need to be half-pixel-exact, i.e. we have unit tests with a certain tolerance. Benoit > this is one kind of trick that driver writers rely on > to optimize such cases, but they can only do so by recognizing the > state+shader fingerprint of that, which is typically application specific. as > a result of these identifiable fingerprints, they can avoid overhead in some > blurs which remove "must be pixel-perfect here" as a requirement. > > this the kind of thing (to over-generalize/-simplify) they do to make blurred > areas in various games and the blur in windows 7 so performant. (i'm told that > win7 uses a rather sub-optimal blur algo, actually, but that because it can be > "fingerprinted" by the driver, windows drivers are highly optimized for it and > so it's very fast in spite of the implementation.) > > the moral here is that it's probably best to present as uniform an appearance > as we can to the drivers (which is a combination of state+shader, not an API) > and let the graphics gurus show us what we actually need ;) > > -- > Aaron J. Seigo > humru othro a kohnu se > GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA EE75 D6B7 2EB1 A7F1 DB43 > > KDE core developer sponsored by Qt Development Frameworks > > >>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe << > > >> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<