Linked by Thom Holwerda on Wed 28th May 2008 19:09 UTC
Thread beginning with comment 316051
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.





Member since:
2005-07-06
Still, OSX and Windows both accelerate more drawing functions than X (or any X-based GUI toolkit) currently does. IMHO they should modernize the available graphic drawing function set and accelerate them where possible. And yes, GTK+ should allow the user to take advantage of Glitz. Even if it was not on by default it should be possible to enable and disable it at runtime.
I remember when the acceleration idea first came out before Windows Vista. The issue that was discussed was this; given the amount of bandwidth, latency, memory on the graphics card and computation power of the said graphics card - do the extra complication one has to go through in terms of designing the necessary subsystems, can it be justified for the apparent (in theory) performance boost?
I think that you need to look at WDDM model, not only now, but the model in the future for version 2 and 3, and the added complication which Microsoft is putting on the hardware vendors. Infact, by the time version 3 of the WDDM roles around, the drivers in itself will almost be like an operating system - where there is threading, resource management, scheduling on the gpu and so forth. If you think that things are bad now, just you wait till that point.
As for Xorg, there was a move a while back to actually put the whole of the Xorg server running ontop of OpenGL; so basically you would have the whole server accelerated. Unlike Microsoft, Xorg doesn't have the luxary of just being able to just jump up and do things; there are more operating systems that use Xorg besides Linux; there is *BSD, OpenSolaris and a few other ones I forgot to mention. What ever is added has to be agnostic to all platforms. With that being said, I don't think the issues are as dire as some people here make out them to be. The issue in alot of cases isn't the right underlying technologies but getting the toolkits to use them. Take GTK+ and Qt, and libxcb for example - and yet, there has been no move from what I see by GTK developers to move to it - which provides improvements over libX11 in terms of latency hiding and so forth.