Ailuros said:...and it stands for what?
You have to read it from left to right correct? Developers...sheesh
j/k
OpenGL guy said:Not to repeat myself, but TANSTAAFL.Ailuros said:There are also ways of getting bandwidth free AA you know
Like the AA on the Kyro. It costs fillrate to activate and transistors to implement, but it doesn't cost bandwidth.
Ailuros said:What makes you think that if you can truly get bandwidth free Supersampling, that Multisampling then would influence bandwidth in any way?
It was just an example anyway. By the way wasn´t Supersampling until recently the "superior" method to antialias?
Multisampling is a more efficient way of doing AA: You don't need to waste time supersampling the textures as better filtering algorithms do a better job anyway. As shader programs get more complex, MSAA will become more important.Sabastian said:Multisampling is cheap/cheat AA of course that is MO.
That was one of the big problems comparing AA on the Radeon 8500 vs the GeForce 4.SSAA does far more work yet does not get recognized with benchmarks......
My point was that you are making a tradeoff someplace. I'd prefer if people gave the tradeoffs being made when talking about something like "bandwidth free AA". OTOH, I wouldn't call the kyro's method bandwidth free because you take a bandwidth hit going through the geometry more than once. Of course, you take this hit with or without AA, but it is still a tradeoff.Thowllly said:OpenGL guy said:Not to repeat myself, but TANSTAAFL.Ailuros said:There are also ways of getting bandwidth free AA you know
Yeah, but he’s not saying 'free AA', he’s saying 'bandwidth free AA'. Like the AA on the Kyro. It costs fillrate to activate and transistors to implement, but it doesn't cost bandwidth.
There's a good example of a tradeoff, thanks(Ok, it's possible the smaller tilesize would cost bandwidth, but they could have made one that used the same tilesize for AA and non AA, so it ‘bandwidth free AA’ could be done.)
That was one of the big problems comparing AA on the Radeon 8500 vs the GeForce 4.
MiNdVelocity said:What are the chances that Nvidia is leaking all these limited availabilty rumors in an attempt to make Ati relax a little bit. I know that it is highly unlikely, but it isn't bad bad idea; however it might backfire and send customers flocking to Ati for early Christmas shopping.
No I am afraid not. Nvidia has nothing to gain at this point by letting a perceived competitor appear superior.
Nvidia representatives declined to comment, citing company policy against commenting on products before they're officially announced.
PVR_Extremist said:I'll stick by those now infamous words from Kristof....
"Nothing is for free in 3D" ?
8)
OpenGL guy said:Ahh, people need to read more Robert Heinlein (I recommend "The Number of the Beast" or "Job: A Comedy of Justice"). Anyway, TANSTAAFL = There Ain't No Such Thing As A Free Lunch.Ailuros said:...and it stands for what?
You have to read it from left to right correct? Developers...sheesh
j/k
Randell said:PVR_Extremist said:I'll stick by those now infamous words from Kristof....
"Nothing is for free in 3D" ?
8)
damn I was going to post that!