Attn Dave: Can I see benched

[CDAZ]

9600 pro running at 500 core and 700 mem and a geforce FX 5800 pro running at the same speeds With apples to apples AA + AF?

 

[Dave Baumann]

Unfortunatly there is no truely equal AF setting between ATI and NVIDIA at the moment.

 

[CDAZ]

I see

Well then I can see what benchmarks would be without AA & AF?

 

[mczak]

Unfortunatly there is no truely equal AF setting between ATI and NVIDIA at the moment.

IIRC 2xAF Quality on Radeon 9[5|6|7|8]00, 2xAF Application on GFFX should be very comparable. That's not much anisotropy unfortunately...

 

[demalion]

Well, there is "0x" AF, True trilinear that's comparable, if you want to count that as an AF setting.

 

[Hellbinder]

well, thats not the core of the issue. The problem here is that ATi and Nvidia have completely different methods of AF implimentation. One is hardware and patented (ATI) the other is Software and can change with every driver release (Nvidia). One uses a *what needs to be covered* approach.. thge other covers everything but hacks up the IQ for everything in order to get speed.

It is literally impossible to do any kind of a true apples to apples comparrison. The best you can hope for is a close approximation of the Eye appeal of teh final outputs of each.

(lest we forget by definition Anistropy does not need to be addressed for every angle and surface)

 

[Rodéric]

A minute I thought you wanted to benchmark Dave... and the whole Beyond3D team

That would be rather funny.

 

[Xmas]

]One is hardware and patented (ATI) the other is Software and can change with every driver release (Nvidia). One uses a *what needs to be covered* approach.. thge other covers everything but hacks up the IQ for everything in order to get speed.

Please, try to find out how AF really works and don't guess to fill the gaps...

I agree with mczak, 2xAF (Quality vs. Application) should be very close. Simple trilinear filtering wold be "1x" AF, btw

 

[Ostsol]

NVidia's "Application" trilinear is actually quite superior to ATI's "Quality". . . Just check Beyond3D's reviews/previews for the screenshots.

 

[Reverend]

]The problem here is that ATi and Nvidia have completely different methods of AF implimentation. One is hardware and patented (ATI) the other is Software and can change with every driver release (Nvidia). One uses a *what needs to be covered* approach.. thge other covers everything but hacks up the IQ for everything in order to get speed.

 

[whql]

NVidia's "Application" trilinear is actually quite superior to ATI's "Quality". . . Just check Beyond3D's reviews/previews for the screenshots.

Who so? Without AF they both appear to be the same when NV is using the Application mode.

 

[Ostsol]

NVidia's "Application" trilinear is actually quite superior to ATI's "Quality". . . Just check Beyond3D's reviews/previews for the screenshots.

Who so? Without AF they both appear to be the same when NV is using the Application mode.

GeforceFX 8x Application
Radeon 9x00 16x Quality

 

[demalion]

Ostsol, you're entitled to your opinion, of course, but people keep saying "no AF".

And the escalation of the colored mip level conflict continues out of control.

 

[Ostsol]

Ostsol, you're entitled to your opinion, of course, but people keep saying "no AF".

Sorry. . . bad habits of not fully reading posts are returning. . .

 

[Xmas]

NVidia's "Application" trilinear is actually quite superior to ATI's "Quality". . . Just check Beyond3D's reviews/previews for the screenshots.

Who so? Without AF they both appear to be the same when NV is using the Application mode.

They actually show very slight differences:
GFFX 5600
R9600

I don't know which one is better. R9600 seems to use the first mip level earlier, but at the end of the tunnel it's just a tad the other way round.

 

[OpenGL guy]

Simple trilinear filtering wold be "1x" AF, btw

Only if trilinear is enabled Remember, AF is a min/mag filter, the mip filter is not specified; "trilinear" normally refers to the min/mag/mip filters all being linear.

 

[CDAZ]

So how do you think

The two cards in question would compare at similar speeds?

 

[Tagrineth]

GeforceFX 8x Application
Radeon 9x00 16x Quality

But at 90º angles, ATi's is hands down superiour.

 

[Tahir2]

Application mode in the NVIDIA drivers is indeed true trilinear but it has an enormous performance hit!

 

[KimB]

GeforceFX 8x Application
Radeon 9x00 16x Quality

But at 90º angles, ATi's is hands down superiour.

Except the colored MIP maps only tell half the story. The other half is texture aliasing. More aggressive LOD is only better if there isn't any additional aliasing.

But ATI's 16x should definitely be superior at horizontal/vertical edges anyway.