PDA

View Full Version : ATI and FBuffer: false publicity ?



divide
02-15-2005, 04:34 AM
As some of you may know, Fbuffer should allow to write unlimited shaders by decomposing implicitly the shader in multiple passes. That would get rid of the limitations of ATIs cards in term of pixel/fragment shaders (especially the texture indirections, limited to 4 even in the very last ATI cards).

I wrote a year ago to ATI to know if FBuffer was implemented (I had a radeon 9800) and they answered me it was on the way.
I wrote back to them a few month ago to know the status, and they said there was no plan anymore to implement this !
What is more, a friend of mine has a x800 and you can read in the specs
"2nd generation F-buffer technology accelerates multi-pass pixel shader programs with unlimited instructions"
(ATI x800 specs) (http://www.ati.com/products/radeonx800/specs.html)

That is actually a lie, seems no-one can take advantage of this, and the card stitch to its native limits...

What do you think about this ?
Does anyone from ATI has a word on this ?

3B
02-15-2005, 06:29 AM
Originally posted by divide:

"2nd generation F-buffer technology accelerates multi-pass pixel shader programs with unlimited instructions"
(ATI x800 specs) (http://www.ati.com/products/radeonx800/specs.html)
I noticed that also...in fact, seeing that bullet point in the X800 specs and still not having any way to use it in my 9800 was one of the factors that convinced me to get an nvidia card for my last upgrade(and probably the next few as well)...

It was especially annoying since I probably wouldn't have gotten the 9800 in the first place if I'd known the feature wasn't actually useable.

Guardian
02-25-2005, 04:58 AM
Indeed the promise of the FBuffer feature made me buy a Radeon 9800 Pro instead of a 9500 or 9700 one ...

So ATI guys ? any comment on that ?

Moreover, when i read the different reviews before buying the board, most of them talked about a full OpenGL 2.0 support ... which is not the case (atm) because for instance it does not support the ARB_texture_non_power_of_two extension ...

Your comments ?

NitroGL
02-25-2005, 09:27 AM
I should point out, that on the x800 it does indeed support the fbuffer in GLSL. I can FAR exceede the indirection count with out forcing software render.

Korval
02-25-2005, 02:59 PM
I should point out, that on the x800 it does indeed support the fbuffer in GLSL. I can FAR exceede the indirection count with out forcing software render.Really? You are the first person on these forums who has reported this. Can you post a shader that does exceed the indirection limitations for others to try out?

NitroGL
02-26-2005, 08:23 PM
Originally posted by Korval:

I should point out, that on the x800 it does indeed support the fbuffer in GLSL. I can FAR exceede the indirection count with out forcing software render.Really? You are the first person on these forums who has reported this. Can you post a shader that does exceed the indirection limitations for others to try out?I don't have one on hand, wouldn't take long to write one though.

Silkut
02-27-2005, 04:41 AM
It's very strange, like divide's link, I read that
latest generations of graphic cards support fBuffer

I don't remember if the HDR Rendering from the NVSDK 8.5 use fBuffer, but my GeForce 5900XT run this application whithout any problems, If NitroGL said that x800 supports fBuffer, maybe he had one, but it's clear that ATI's pub is not exactly the reality (if they said that the 9800pro support fBuffer..)

Divide, you wrote to them and they said there were no plan to implement fBuffer on X800 ????
WTF :confused:

divide
02-27-2005, 12:39 PM
"Currently there is no plan for implementation but if there is a change in plans it will be posted at www.ati.com" (http://www.ati.com") is what they answered in their last email, dated January 18th 2005.

Silkut
02-27-2005, 03:37 PM
That's incredible, what about the X850 ?