ATi is ch**t**g in Filtering

[digitalwanderer]

I don't think the people knowledgeable were at the E3, most of them should be PR mans.

Uhm, wouldn't the PR people be who would address this from ATi? :|

That's why I figured I didn't get any of my e-mails returned from ATi PR people last week at least, either that or they've secretely hated me all along and just chose now to let me know about it.

 

[Dave Baumann]

Actually, you'd be surprised at how many of the tech types go to things like E3 - its a very good place to see whats coming up if you've not already seen it and you can chat to developers to find out what they are doing / where they are going. Its not all about the booth babes!

 

[Chisholm]

Its not all about the booth babes!

Is it not?

 

[Richthofen]

3. I think releasing something of this magnitude on a weekend is not in the best interests of anyone involved as it seems to have caused a great deal of commotion/speculation that might have been avoided if you had waited for an appropriate response. Then, if the response from ATI supports your news item, THEN paste it all over the internet if that is your intent. If I were with ATI, I would almost feel like I had been "set up" simply from a timing standpoint....

How do you think Nvidia felt over the last months? While some accusations were true others were not. But no one - especially most hardware websites - cared.
Why should ATI be treated any different?
They had more than enough time since the first computerbase article concerning a Radeon9600. No reaction.
Well now it's their task to talk about this issue. But i am sure they will keep on stating what kind of first class filtering their products have like they did in their pdf documents related to R420.
I must have missed something. Since when did ATI have had world class filtering?
Since R200 their filtering has never been at the level of their competition.
AA is a different story but that's not the iissue here.

 

[Stryyder]

I don't think the people knowledgeable were at the E3, most of them should be PR mans.

Yes why should they have knowledgable people to work with the hundreds of developers there or to support the products the developers are using

 

[DrawPrim]

No, but a driver sure could.

Hmm, there is no reason to detect the differences between the mipmaps. The differences are usually small. But if you want to detect if someone is using couloured mipmaps, in this case this detection is very useful.

Mips should be similar; however they could differ enough that they would want to detect the differences and adjust the filtering aggressiveness. I've already said that the MS conformance tests will use wildly different mips levels (vertical bars, horizontal bars, etc) to detect filtering problems. My guess is that if you created mips levels that aren't colored but are just very different, you'd see the same behavior.

 

[Mintmaster]

Thanks a lot Lars, that would explain a lot.

"brilinear" is pretty easy to see even without mip-map colouring if you have the right texture. I remember low res road lane markers in racing games looking very ugly without anisotropic filtering, because the low angle made the lines blur and disappear in a hurry. You can see the filtering quite clearly.


I think this is a good way to do brilinear, though, because some algorithms rely on a smooth, linear transition across mip-maps.

One method of doing caustics in the vertex shader takes advantage of the fact that mip-map selection is based on quad area size, so the light intensity at the bottom of a pool is determined this way. (sorry for the unclear explanation...).

Another need for smooth mip-map filtering is zeckensack's suggestion of getting a gradient based value into R300/R420.


However, it is very misleading, and many reviewers disabled trilinear optimizations on NV40. It should be an option, as everyone has mentioned. Whether they make it default or not is up to them, but if NVidia made it an option, surely ATI can.

 

[digitalwanderer]

Actually, you'd be surprised at how many of the tech types go to things like E3 - its a very good place to see whats coming up if you've not already seen it and you can chat to developers to find out what they are doing / where they are going. Its not all about the booth babes!

Actually I quite believe that, it sounds like a dream opportunity to truly geek out with some like-minded people. 8)

 

[Heathen]

So judges, do we have a conclusion yet?

Dunno, but the sooner we gat an official response from ATi the better.

 

[Hellbinder]

I would imagine that if it was something easily addressed outside of a PR marketing move that OpenGL guy or Sireric would have already commented on it.

Which does not bode very well on the issue in my mind.

(Of course perhaps they are on vacation or soemthing)

 

[Evildeus]

I don't think the people knowledgeable were at the E3, most of them should be PR mans.

Yes why should they have knowledgable people to work with the hundreds of developers there or to support the products the developers are using

Yeah they need the knoledgeable guys to put the card in the PC

 

[Evildeus]

Actually, you'd be surprised at how many of the tech types go to things like E3 - its a very good place to see whats coming up if you've not already seen it and you can chat to developers to find out what they are doing / where they are going. Its not all about the booth babes!

Yeah, but that doesn't mean that all Ati was @ E3, or does it?

 

[no_way]

As most "applications developers" are not even caring about Anisotropic or other methods of Filtering or even have controls for it in their software.

We'll just have to disagree. Not having controls in application configuration doesnt mean that developer doesnt care. The "setup section" you usually see in PC titles is there as just a necessary evil, you dont see those much in console titles. This doesnt mean that app isnt calling for specific behaviour by rendering subsystem.

It is rare that developers even think down that road.

A guy developing a 3d rendering engine and not having a grasp on texture filtering concepts, with some clue on which filtering options are appropriate for which textures is simply a wrong guy in the wrong place. As for middleware renderers users: simply dont use crap.

Which is why Nvidia and ATi have control panel settings for AF in the first place.

I always had the impression that those are present simply to improve legacy application look.

Now in the case where the application is trying to use a specific filtering method in a predetermined way then yes the IHV should leave it alone again *IF* set to application in the control panel.

Application is ALWAYS trying to use a specific filtering method, whether API default or something else doesnt matter. AFAIK theres no way to ask for "gimme any random filtering on this texture" option neither in Ogl nor DirectX

 

[Evildeus]

Thanks Quasar, that's what i wanted to know. Interesting, thanks once more.

Yes of course, the question was more like, why did you make those comparisons? But i realise that's a bit dumb, if you found some differencies, sure you would investigate more and more to see what's going on. Sorry for asking for something so obvious

There are no dumb questions - only silly answers.
Initially what struck me were benchmarks, were both (X800 and nV40) were pretty close in normal-mode - with a huge lead for nV40 with AA and an even bigger advantage for ATi with AF on. With both turned on, the lead was on ATi's side despite the fact that popular belief was, that both were mainly bandwidth limited.
Then there was this PCGH-article where they measured a 20% increase of trilinear fillrate over it's theoretical peak with X800 (reproducible on both RV350/360 and R200) and of course the ColorMip-Thingy with CoD, which we noticed one week ago in our news item.

Then we started to dig.

 

[Big Bertha EA]

3. I think releasing something of this magnitude on a weekend is not in the best interests of anyone involved as it seems to have caused a great deal of commotion/speculation that might have been avoided if you had waited for an appropriate response. Then, if the response from ATI supports your news item, THEN paste it all over the internet if that is your intent. If I were with ATI, I would almost feel like I had been "set up" simply from a timing standpoint....

1. How do you think Nvidia felt over the last months? While some accusations were true others were not. But no one - especially most hardware websites - cared.

2. Why should ATI be treated any different?

1. I said the same thing about nVidia when numerous issues surfaced regarding FX drivers. It is my opinion that the whole 3D hardware industry has become "micro-scrutinized" for lack of a better term. I am all for honest and informative reviews, but when you have hardware editors jumping through hoops to find even the slightest "anomaly" in drivers or output, the consensus view these days has become "guilty until proven innocent". I feel that the ONLY fair way to deal with issues of this nature are to exhaust every conceivable means of contacting the company suspected before splattering speculation all over the 'net. To me, it is unfair to any company to do anything less.

2. I NEVER said ATI should be treated differently.

 

[Pete]

However, it is very misleading, and many reviewers disabled trilinear optimizations on NV40. It should be an option, as everyone has mentioned. Whether they make it default or not is up to them, but if NVidia made it an option, surely ATI can.

I don't want to read too much into it, but maybe this is why nV "disabled" tri with the 61.11s? It would be somewhat strange for nV to keep quiet about this if they'd discovered it, though.

Anyway, if one theory is that ATi is detecting and then optimizing for specific texture patterns, wouldn't it be possible to slap a bunch of test textures onto a custom UT2K3/SS:SE/whatever map to dis/prove that theory?

 

[Colourless]

For whatever reason, I sometime 'forget' about things I've already done. For instance, it is possible for me to modify my Direct3D Intercepter DLL that i use with D3DBench (and also R300 Ruby Rap) to modify textures

 

[Hanners]

Quite possibly it's not purely intended to hide what's going on with coloured mip levels, but could also be intended to be used with any texture that has mip levels that are quite different.

My preliminary understanding is that this is probably closest to the case. One may want to "think historical" as well.

We're talking Quake 3 a la Radeon 8500 era here, right?

 

[Dave Baumann]

No, before that. Not necessarily ATI either.

 

[CarstenS]

Quite possibly it's not purely intended to hide what's going on with coloured mip levels, but could also be intended to be used with any texture that has mip levels that are quite different.

My preliminary understanding is that this is probably closest to the case. One may want to "think historical" as well.

We're talking Quake 3 a la Radeon 8500 era here, right?

Well then let's hope the next driver release does fix this issue with no performance drop.