Guess ATi forgot to test their demo on FX boards

[Ante P]

 

[Doomtrooper]

5800 Ultra I assume...

Nice colour blending though

 

[Ante P]

5800 Ultra I assume...

Nice colour blending though

The Bear demo is even worse, it looks like going outside a clip place and runs at 0.5 fps
car demo is just black

the old 8500 nature demo works fine though and runs sweet even with 4xs FSAA and 8x aniso

 

[pocketmoon_]

! Pipe dreams looks absolutly fine on my Fx 5800 det 44.03

 

[Ante P]

! Pipe dreams looks absolutly fine on my Fx 5800 det 44.03

using 44.10 here

guess I'll have to wait for the next update

 

[Althornin]

beyond the obvious, these demos were out long before the GGFX's were....

Also, mayhap some of the problems have to do with inability to render to a FP target?

 

[Doomtrooper]

Whats with the 44.03/44.10 drivers, you would think the 44.10 drivers would be the best for any game/demo since they stay at full precision.

 

[Ante P]

Whats with the 44.03/44.10 drivers, you would think the 44.10 drivers would be the best for any game/demo since they stay at full precision.

beats me :?

 

[Pete]

Far out, man.

 

[AlphaWolf]

Ya but does it beat the ATI score? that's all that really matters.

 

[KimB]

Ya but does it beat the ATI score? that's all that really matters.

Not really. You can be certain that those ATI demos used no partial precision (even though they probably could in many places in the demos), since the R3xx architecture always works at 24-bit FP.

The FX architecture depends on a good number of the shader instructions within each shader using partial precision for good performance.

 

[Althornin]

partial precision for good performance.

eh, depends on your definition of good.
From the tests we've been seeing, it looks like even partial precision PS2.0 is much slower than ATI.

 

[Arun]

partial precision for good performance.

eh, depends on your definition of good.
From the tests we've been seeing, it looks like even partial precision PS2.0 is much slower than ATI.

Well, from my understanding, NV30/NV31/NV34 is slower at FP16 than the R300 at FP24 while the NV35 at FP16 is faster than the R300 ( no idea against the R350 ) at FP24.

Could be wrong, of course.


Uttar

 

[Bjorn]

Well, from my understanding, NV30/NV31/NV34 is slower at FP16 than the R300 at FP24 while the NV35 at FP16 is faster than the R300 ( no idea against the R350 ) at FP24.

Could be wrong, of course.

Uttar

AFAIK, Carmack is the only one that has confirmed this. And that's only for Doom3.

 

[K.I.L.E.R]

Ante P, are you showing us the world in your eyes when you are high/drunk?

 

[Ante P]

Ante P, are you showing us the world in your eyes when you are high/drunk?

I've told my FX so many times "STOP DOING ACID MAN" it just won't listen
my patience is running out

 

[Landcruiser]

Try the chimp demo.

It looks like Rayman

 

[Hyp-X]

Pipe demo works with the 44.03 on FX5600.

The image quality looks identical despite that this driver forces FP16.

Believe me there's is ample time to analize image quality as it barely reaches 2 fps. (yuck!)

 

[Hyp-X]

Try the chimp demo.

Well it has a face, hands, feet, belly button, so do we miss anything?

(Butterflies also look funny without their wings...)

 

[martrox]

Try the chimp demo.

Well it has a face, hands, feet, belly button, so do we miss anything?

(Butterflies also look funny without their wings...)

ACK! Those damned nVidia videocards are pulling the wings off butterflies....... Those bastards!