chip has a review up

[Doomtrooper]

L233,

I understand yet in their reviews in the past they have included screen shots and it amazes me they wouldn't show a 6X FSAA shot

 

[Nagorak]

the numbers are not really compareable because they used quincunx aa and 8x aniotropic on the gf4600 while using 6x multisample and 16x anisotropic on the r9700 but still impressive

WTF, they used Quincunx?????!! The glorified blur filter?!!!!

How about a comparison of things that result in the SAME image quality!

For that matter, how about comparing the cards with the same settings...

And what's with GL Mark? Is that benchmark totally made for Nvidia cards or what?

 

[Ailuros]

The author claims that he supposedly used the maximum sampling rate of each card. Where and why he found Quincunx to apply more samples than 4xOGMS is one of those riddles.

 

[David G.]

Anti-aliasing and Anisotropic filtering will always end up being apples to oranges comparisons due to the very different implementations - unless great care is taken to create a level playing field.

Then again comparing 16x Anisotropic + 6x MS-AA with 8x Anisotropic + 4x MS-AA is a bit silly and results should come with a huge warning sign

K-

As a matter of fact ... from what I know Quincux - 2AA + some blurring efect .


6xAA vs 2xAA !!!! ????

16x Aniso vs 8x aniso ?!?!?!


That's not a "pourely done review" .. it's more like and nVIDIA marketing document .

 

[Nagorak]

The author claims that he supposedly used the maximum sampling rate of each card. Where and why he found Quincunx to apply more samples than 4xOGMS is one of those riddles.

The point is what he did doesn't even make sense anyway. If you want to compare card's speed you don't compare them at different settings. I mean if the R9700 offers higher quality images that's great, but what purpose is there to compare that to a lower quality image on the GF4.

I could see if he put screen shots of max quality and pointed out that the R9700 was way higher but still the same speed or something...

 

[Andergum]

Terrible review. Not even worth looking at.

Hopefully we'll see some well done reviews done by competent people in the next few days that will at least compare the cards at the same settings.

 

[tEd]

they also broke the nda again!!!!!!

 

[gkar1]

The author claims that he supposedly used the maximum sampling rate of each card. Where and why he found Quincunx to apply more samples than 4xOGMS is one of those riddles.

The same way some site said the 8500 didn't do "true anisotropic filtering", and excluded it from a comparison shootout.

 

[multigl2]

i dont understand, the r200 does anisotropic filtering as correct as the GF3 does; if they wanted to do maximum settings on all cards, they didn't meet their desires. Obviously, the Radeon9700 is running max (6x MSAA, with 16x Trilinear Anisotropic) and the GF4 isn't. For the fairest comparison, both should do max anisotropy (16x Trilinear on both?? ) and 4x MSAA.

 

[nggalai]

a usual chip review i would say

Exactly. Yet again a sucky Chip review. What's more, it's already the fifth time the guys broke an NDA. ATi and OEMs such as Sapphire are not amused, let me tell you.

ta,
-Sascha.rb

 

[Kristof]

I can't remember did they not have 2 Quincux modes... hmmm now that I think of it the standard Quincux pattern does only have 2 true sample points and borrows 3 more. I think on GF4 they just moved the 2 samples around a bit...

I think we have pretty much agreed that the comparison is very much invalid.

K-

 

[ram]

fifth time the guys broke an NDA. ATi and OEMs such as Sapphire are not amused, let me tell you.

There won't be any consequences. ATI will give the boards the next time again. So did Matrox.

 

[nggalai]

Still sad state of affairs. Too bad companies need exposure this badly that larger media can more or less fare how they want.

ta,
-idealist.rb

 

[Bambers]

By counting the 3 blur samples as separate ones you can get quincux as a 5sample AA. Although on that you could use 4x9tap as 9 samples and so the most sample AA on the card. That would have hit the gf4s fps too badly for them though