Anisotropic Filtering & Multi-texturing on GF3/4 a no-go

I'm not sure if you guys missed it or not....

But in the article, I stated that I ran the same type of experiment on GL_EXT_reme, and I got *virtually* the same exact fill-rate results...It was off by like 20 MTex/sec, but it was totally consistent with 3DMark2001.
 
The fill rate test is not affected by aniso on the 8500.

When you force aniso in d3d does it force on trilinear too?
 
Bambers said:
The fill rate test is not affected by aniso on the 8500.

When you force aniso in d3d does it force on trilinear too?
Click to expand...

The R200 core is incapable of aniso + trilinear.
 
did some tests on that matter!

gf3ti500(240/550)
3dmark fillrate test
si mu
wo 756.2 1612.2
2tab 490.8 960.6
4tab 490.8 960.6
8tab 490.8 960.6

gf4ti4600(240/550)
3dmark fillrate test
si mu
wo 869.2 1872.7
2tab 493.6 494.1
4tab 493.6 493.6
8tab 493.6 493.0

hopefully only a driver problem!?
 
i discovered the same problem here when comparing my GF3Ti500 with my GF4Ti4400.
The fillrate of the GF4 in Multi-Texturing when aniso is enabled drops to the singletexturing scores.
Seems to me that there is a huge bug. I still have the problem with the new 28.90 driver leaks from guru3d but in opengl (quake3) my aniso performance got a huge performance boost.
Maybe you wanna try those new driver leaks on your system.
 
Bambers said:
Tagrineth said:
Bambers said:
The fill rate test is not affected by aniso on the 8500.

When you force aniso in d3d does it force on trilinear too?
Click to expand...

The R200 core is incapable of aniso + trilinear.
Click to expand...

I meant the gf3/4 with the question :)
Click to expand...

No, anisotropic doesn't force linear mipfilter on GeForce cards.
 
Old post but can someone answer me on this:

So can a GF3 Ti500 let say outdo the GF4 in general performance when doing max AF?

How about 4x MSAA and 8X AF with trilinear samples?

Anywhere on the net to see the results?

Seems to me that the GF4 series will become fill rate limited when doing high IQ gaming. No wonder people couln't belive my GF3 Ti200 frame rates with 4x AA and 8x AF using tri samples, abeit 16bit with 32bit textures.
 
noko said:
So can a GF3 Ti500 let say outdo the GF4 in general performance when doing max AF?
Click to expand...
Yes it can, but only if part of the textures use only bilinear filtering (eg. lightmaps).

MSAA is usually faster on GF4.
 
The GF4 will win in bandwidth limited scenarios (including MSAA) - it has a more efficent memory controller.

Actually 2x and Quincunx AA even uses less bandwidth on GF4 as it doesn't downsample the buffers to the memory.
 
You must log in or register to reply here.
Back
Top