9700 benchmarks

[KimB]

3dmarks != real gaming perfromance

Never has, never will. Only use 3dmarks to gauge and verify your own system. Its useless for anything else.

I'm not even sure it's good for that, since it's such an inconsistent benchmark (Good benchmarks run within a fraction of a frame each run...a 3DMark2k1 score can change by hundreds of points between runs).

And, of course, given that we've seen absolutely no game that follows 3DMark2001's performance ratings, it is of even less use of, for example, judging driver quality.

The only thing I do kind of like are the completely synthetic tests. After all, since the game tests essentially are synthetic, they're only pretending to be realistic, they are of absolutely no use. At least in the synthetic tests we can see what part of the video card is being stressed.

 

[Doomtrooper]

Since we've seen the 9700 numbers and graphs I was looking for some screen shots for IQ comparison. I noticed this from Hothardware review(the nice thing about posting screen shots when talking graphic cards..what a concept) was the default lod on these FSAA screen shots..no anisotripic filtering was used or mentioned here in these tests.

9700 shot 4XFSAA:

Geforce 4 Ti4600 4X FSAA

Matrox Parhelia 16xFAA

 

[Entropy]

The Parhelia looks damn good.

Entropy

 

[KimB]

The Parhelia looks damn good.

Entropy

Until you look at the helicopter blades.

Anyway, in these first few shots, the methods used are almost indistinguishable between the GeForce4 and Radeon 9700. Hopefully we'll have some more in-depth comparisons soon.

 

[Joe DeFuria]

For sure, the R 9700 has a higher LOD than both the Parhelia and GeForce4. (Is that in D3D or OpenGL?)

It's not hard to spot either, though the "easiest" way to tell is to look at the stars in the American Flag....

 

[andypski]

The Parhelia looks damn good.

Yup, apart from the complete lack of antialiasing on the helicopter blades.

The AA on the rest of the image is very nice, however.

 

[Colourless]

Apparent Higher textures quality on R3700 aside, can someone explain why the grass at the bottom of the images behind the chopper looks different in all three shots.

 

[Doomtrooper]

Comanche is a Direct3D game, looking at the ground textures and the trees the 9700 is a much better LOD setting. The trees in the distance are not blurred..use a picture viewer and switch back and forth or overlay....it stands out bigtime.

 

[Joe DeFuria]

Anyway, in these first few shots, the methods used are almost indistinguishable between the GeForce4 and Radeon 9700. Hopefully we'll have some more in-depth comparisons soon.

I disagree. I would say in those images, the ATI AA is noticably better than the GeForce4.

I would like to see GeForce4 4Xs AA vs. ATI's 4X AA.

can someone explain why the grass at the bottom of the images behind the chopper looks different in all three shots.

Perhaps that engine "randomly" places those clumps of grass? within a defined area?

 

[KimB]

Comanche is a Direct3D game, looking at the ground textures and the trees the 9700 is a much better LOD setting. The trees in the distance are not blurred..use a picture viewer and switch back and forth or overlay....it stands out bigtime.

More aggressive LOD doesn't always mean better LOD. However, given that the 9700 is capable of (apparently proper this time!) 16-degree anisotropic, it definitely should be able to enable a higher LOD setting without introducing higher amount of aliasing. I would still like to see some comparisons between the implementations when all cards are using the same degree of anisotropic, however (such as MIP map selection, both method and distances...).

 

[Doomtrooper]

Comanche is a Direct3D game, looking at the ground textures and the trees the 9700 is a much better LOD setting. The trees in the distance are not blurred..use a picture viewer and switch back and forth or overlay....it stands out bigtime.

More aggressive LOD doesn't always mean better LOD. However, given that the 9700 is capable of (apparently proper this time!) 16-degree anisotropic, it definitely should be able to enable a higher LOD setting without introducing higher amount of aliasing. I would still like to see some comparisons between the implementations when all cards are using the same degree of anisotropic, however (such as MIP map selection, both method and distances...).

Yes we've been down this road before Chalnoth..I'm not stating aggressive LOD..

 

[KimB]

Yes, I noticed that when I finally went back and opened the pictures in different windows.

Regardless, haven't I posted before that in a static screenshot, it's easy to produce a shot that looks great, but has tremendous aliasing in motion? This is why I would really like to see some comparisons using the same degree of anisotropic. If the Radeon 9700 has a more aggressive LOD setting than the GeForce4 under the same degree of anisotropic, personally, it looks very bleak for the aliasing aspect of the card.

What I would really like to see is a screenshot from a program/situation that is specifically designed to stress texture aliasing.

 

[Joe DeFuria]

If the Radeon 9700 has a more aggressive LOD setting than the GeForce4 under the same degree of anisotropic, personally, it looks very bleak for the aliasing aspect of the card.

And this is because?

The GeForce4 has the only true and correct level of LOD based on the only true and correct aniso algoithm?

In any case, if the LOD is "too aggressive", then simply adjust the texture sharpness slider in ATI's control panel to your liking.

I thought you of all people would be applauding ATI for being "Anti-Cheaters"...using higher relative LOD's that could only serve to lower their own benchmark scores...

 

[KimB]

The GeForce4 has the only true and correct level of LOD based on the only true and correct aniso algoithm?

Which is why I said, "looks bleak." It is certainly fundamentally possible to produce an image at the same degree of anisotropic with less aliasing...it's just unlikely, as it would almost certainly require more fillrate, or, at the very least, significantly more math.

In any case, if the LOD is "too aggressive", then simply adjust the texture sharpness slider in ATI's control panel to your liking.

I'd rather have a default sharpness that is designed to minimize aliasing.

I thought you of all people would be applauding ATI for being "Anti-Cheaters"...using higher relative LOD's that could only serve to lower their own benchmark scores...

I see it another way. Static screenshots of aggressive LOD's almost always look superior in a static image. They only start to become visibly poor when in motion (or in a scene that specifically stresses aliasing...).

 

[OpenGL guy]

I'd rather have a default sharpness that is designed to minimize aliasing.

Next driver release I'll be sure to allow you to be able to force use of 1x1 mipmaps only, Chalnoth.

That should take care of any texture aliasing

What evidence do you have, other than one screenshot, that the 9700 will have aliasing problems? Maybe you can have sharper images that aren't aliased.

 

[Joe DeFuria]

Yes, Chalnoth, I fully understand the implications of more aggressive LOD and aliasing when in motion.

My (I thought obvious) point, was that when ATI has the more aggressive LOD, your concern is all about the ATI part and aliasing artifacts...and when ATI has the lower LOD, your concern is about the ATI part all about "fair" benchmark scores...

You always seem to be "concerned" about ATI...either "cheating" with FPS scores...or "cheating" with image quality. (Better static image quality at the expense of aliasing in motion...)

 

[Johnny Rotten]

The antialiasing quality between the 9700 and the 4600 are essentially a dead wash (though the GF4 seems do a better job cleaning up the tent support lines. At least in that shot). The 9700 though does seem to have a noticeably higher LOD.

Mipmap boundaries are odd looking to my eye in both the 9700 and 4600 shots.

 

[KimB]

What evidence do you have, other than one screenshot, that the 9700 will have aliasing problems? Maybe you can have sharper images that aren't aliased.

I'm just attempting to say that it's something to look for, and given what I've seen of ATI's hardware in the past, LOD has always been set too high. I don't see why the 9700 would be any different.

Additionally, do we even know for certain what settings were used for anisotropic in those shots? It seems to me that the only possible thing that could account for such a drastic difference in texture clarity, other than a massive increase in aliasing, would be that anisotropic was disabled entirely on the GeForce4.

 

[Joe DeFuria]

I'm just attempting to say that it's something to look for, and given what I've seen of ATI's hardware in the past, LOD has always been set too high.

This coming from someone who doesn't have (and claims to have never had) past ATI products?

It seems to me that the only possible thing that could account for such a drastic difference in texture clarity, other than a massive increase in aliasing, would be that anisotropic was disabled entirely on the GeForce4.

I offer another possibility, as far fetched as it might seem to you:

That one possible thing that could account for the difference in texture quality, is that GeForce's is not aggressive enough. GeForce's could be more aggressive, and not exhibit significant aliasing, but it is lowered to help boost performance

Just another "possibility".....

Having said all that it would be nice now that the rush to get the "reviews with benchmarks" out the door is over, if someone would do a more "playability" testing and report on it. To answer things like "does the default LOD seem to aggressive?" Subjective views on the different AA modes, etc...

 

[OpenGL guy]

Additionally, do we even know for certain what settings were used for anisotropic in those shots? It seems to me that the only possible thing that could account for such a drastic difference in texture clarity, other than a massive increase in aliasing, would be that anisotropic was disabled entirely on the GeForce4.

I thought anisotropic filtering was disabled on all the shots.