[Hellbinder]

I opened up a discussion on this at Rage3d.

http://www.rage3d.com/board/showthre...eadid=33685988

Here is my opening statement.

It seems that most people still do not know the truth about the Quak issue.

yes Ati was using application detection, but the *optimization* had been in no less than 7 previeous driver prior to the release of the 8500. the problem was that the exsisting optimization caused an issue on the 8500 wich ATi did not apparently notice. However Nvidia Knew about it and provided the info to [H] for exposure as a *cheat*. Which everyone still repeats to this day. Is application Detection bad?? Only if it intentionally affects the game in an improper way, or intentionally changes the Dynamics of what the application is trying to do. Technically, this is a similar issue to the 3dmark Nvidia scandal. However the difference being. The application detection was present in the driver for a long time, and never affected the game adversly. The Issue was addressed inside of 3 weeks with no performance loss. Later the Quake scores have increased steadily over time.

What is even worse and Doomtrooper please help me out here with a link. When [H] posted their Radeon Vs GF2 article a year before it was apparent that Nvidia had reduced texture Quality in the ENTIRE GAME to boost performance. But No one called it a cheat. ATi's mistake only affected 5 textures in the entire game. But becuase Kyle is best friends with Brian Burke at Nvidia they have a history of Being Nvidia little dirty work boys while glossing over any inconsistancy with Nvidia.

They have gotten better recently i must admit. But every once in a while they still revert to Nvidias little Whore mode. No i dont mean that they have to favor ATi all the time.. Be fair. Its just that when they suddely revert to little Nvidiots it is sinkeningly obvious.

[the floor is now open to anyone with insider info, pertinent info or opinions on the matter]

Here is Doomtroopers kind linkage on the matter

BTW the author of this is not me, but someone here on Rage3D:

http://www.geocities.com/radtruth/p1.htm

Kyles Joy

http://www.hardocp.com/article.html?art=MTEx

The Irony-The quake Reference is in this driver

http://www.hardocp.com/article.html?art=MTEzLDM=

Quack looks good compared to this mess

http://www.3dvelocity.com/reviews/gffx5800u/gffx_5.htm

the Floor is now open.

[RussSchultz]

May as well play the part I'm labeled. I'll paint a target on my chest:

5 specific textures sounds suspect to me (that it was an intentional act at some point and not a bug).

Unless, of course, somebody can come up with some reason why those 5 were fubared and no other ones weren't. (Like some commonality they shared--apart from increasing the scores--like size, format, useage, etc that no other textures shared)

EDIT: And i'm not going to discuss anything else. No motives, no moral relativism, no nothing. My complete discourse will center on these textures and why they were screwed up and others weren't. If you're talking to me, please don't talk about anything else. :P

[Hellbinder]

Right Russ.. It is clear that it did affect the game. 5 textures is a little fishy to. Its simply interesting as you can see in the link doom provided the referecnce to Quake was also in the drivers tested with the origional Radeon. Which of course looked fantastic.

The issue wtih that GF2/Radeon article is that it is clear that Nvidia reduced the Quality in the entire game. Obviously to increase scores. Yet why was that never questioned?

It also raises the point that ATi is usually referenced as the cheater, becuase of Quak... But.. The Evidence seems to be that Nvidia has been Cheating for a much longer time and it still continues to this day.

[MuFu]

Originally Posted by Hellbinder[CE

]It also raises the point that ATi is usually referenced as the cheater...

I honestly don't think that is an opinion shared by the majority.

MuFu.

[Tagrineth]

Did you actually read that? It WAS an intentional optimisation by ATi.

the problem was that the exsisting optimization caused an issue on the 8500 wich ATi did not apparently notice.

In other words, it worked great on the Radeon R6, but caused problems on R200.

Is it that hard to understand?!

[Hellbinder]

The thing is even if you are calling reducing texture detail a little *Cheating*.. then why was the Gf2 not cheating its ass off when it went up against the origional Radeon? why was their low res S3TC hack not cheating? why is their Reduced Quality AA and AF not considered Cheating??? (especially their recent AF)

If the measure of cheating based on Texture detail, or other types of things.. why have the rules changed???

Nvidias entire Nv3x line with the recent drivers should be considered Cheating right?? lower PRecision Quality forced by the Driver regardless of the applications desire and thats just one area.

You simply CAN NOT call what Ati did a *cheat* if you give Nvidia a pass on their Reduced Quality settings. which has been a Constant pattern for them every time they want to win a benchmark for 3 straight years.

[RussSchultz]

Are you talking to me? Remember, I'm not discussing relativism. I don't care if NVIDIA is cheating their butts off at the moment, or for all eternity.

Was the 5 specific textures that were missampled intentional?
Did that positively affect the scores?
Was this circumvented by changing the name of the executable?

Or is that not what we're talking about here?

[OpenGL guy]

Originally Posted by RussSchultz

5 specific textures sounds suspect to me (that it was an intentional act at some point and not a bug).

Unless, of course, somebody can come up with some reason why those 5 were fubared and no other ones weren't. (Like some commonality they shared--apart from increasing the scores--like size, format, useage, etc that no other textures shared)

Why don't you do the leg work since your the one saying it's a cheat? You should be able to get most of the information quite easily.

Does your home have a basement, Russ? If so, I bet you could hide lots of terrorists and weapons of mass destruction there.

Don't worry! I'm not implying anything, just that you have a basement in your house. I mean, no one would ever think that you might hide terrorists in your basement, but who knows? I haven't looked in your basement lately. (And if you don't have a basement, and don't get what I am saying, don't bother to reply.)

[Hellbinder]

Are you talking to me? Remember, I'm not discussing relativism. I don't care if NVIDIA is cheating their butts off at the moment, or for all eternity

No i was just postulating out loud. Not addressing anyone specifically. It is simply that if doling *X* is considered cheating then the practice of *X* should be considered cheating by everyone.

If it is decided that *X* is not cheating but something else.. then that same rule needs to get passed around to everyone.

Was Reducing the Detail levels in Quake by ATi *Cheating* or not. If the Answer is Yes.. then based on that precident then where does this wormhole really lead?

[Tahir2]

It leads to ATI having egg on their face.
But that is not as important as what ATI does next. It has two options, continue with this cheat/bug or do something about it.

Thankfully ATI chose the latter.

Reference to NVIDIA - NVIDIA seem to be doing the opposite and are in fact getting caught out so frequently that soon no one but the most diehard NVIDIA fan will trust them.

It takes years to gain trust and just one act to ruin it all. Whether ATI meant it or not, they have gained back the trust from many people, others might do well to learn from that.

[RussSchultz]

OpenGl_Guy:

So, what you're saying, is that if you can't refute the facts, lets attack the person?

And no, I don't have a basement, and yes, I do know what you're talking about. And no, I'd rather not drag that same discussion into another thread.

Hellbinder wanted to have a definitive thread on the truth of what it was. Using simply the information that Hellbinder supplied (which I'll stipulate is all true, because I really have no clue if it is or not and don't care enough to look anything up) I'm playing the dutiful role of nvidiot and delving into the truth of what quake/quack was/is. Whether any other company is doing really has no bearing on what the truth of quake/quack was/is.

You've actually got access to the source. You're as close to the truth as any of us will get. What was it about? Through what manner did it exist through so many generations of drivers to finally end up causing the result with the new chip? Why only those few textures? Spill the beans, and those beans will be the truth forever more.

Or is simply asking the questions --when prompted to ask-- inflammatory and enough to label one with bias?

[OpenGL guy]

Originally Posted by RussSchultz

OpenGl_Guy:
So, what you're saying, is that if you can't refute the facts, lets attack the person?

Where did you get this from? You were the one asking if there were any similarities between the textures that had the problem. I suggested that you do the leg work and find out for yourself. You'll learn more this way. I don't even know the answer.

And no, I don't have a basement, and yes, I do know what you're talking about. And no, I'd rather not drag that same discussion into another thread.

You're right, I should have made that comment in the other thread.

You've actually got access to the source. You're as close to the truth as any of us will get. What was it about? Through what manner did it exist through so many generations of drivers to finally end up causing the result with the new chip? Why only those few textures? Spill the beans, and those beans will be the truth forever more.

I don't have access to the code in question. For one, the problem was in the OpenGL driver. Two, that was before I joined the company. Third, I don't care about the issue anymore.

[Tahir2]

Originally Posted by RussSchultz

So, what you're saying, is that if you can't refute the facts, lets attack the person?

Unfortunately that is exactly what some people have done in the past, present and future - it's a danger clear and present.

OK me stop now....

Edit: Sorry one more since I can't resist, 'Hellbinder you want the truth? You can't handle the truth!'

[RussSchultz]

Ahhh. You don't care about the issue anymore.

(psst. I really didn't either--at least not enough to actually look anything up. Its ancient history and I'm not sure why it keeps getting brought up. But lets keep that between me and you so I can continue to egg Hellbinder on )


But, doing my duty, Hellbinder: You seem to be the most knowledgable person here. Do you know which 5 textures these were so that I can further investigate (and learn something!) about how they may or may not be related?

[Hellbinder]

yes i do.. Not the specific names.. but it was the floor textures.. (if i remeber correctly)

The reason this keeps getting brought up is obvious. At least to me. The Quak issue is *MINUSCALE* compared to the obvious Repeated violations by Nvidia. Yet, Quak is the only issue ever referenced BECAUSE no one has ever pointed out that Nvidia is the Biggest Violator by a mile. Wether this is due to a doubble standard, or fear of retaliation, or something else I have no idea. But, if messing with IQ for speed is Cheating.. then the nature of benchmarks and who really won them etc.. from the last 3 years should all be reexamined.

[Doomtrooper]

I would like to go on record that I have a basement (not quite sure if that is good or not )

I certainly believe ATI was optimizing(and it looked great on the Radeon 64) I had one and image quality was so superior to my GTS, especially with S3TC. In fact reading the old reviews that is the one thing that was consistent on the Radeon, but like Tagrineth stated what 'optimization' was working good on the R6 didn't work so great on the R200.

Now if Kyle would have got this 'quackifier' from Nvidia one year earlier with this Radeon 32 meg review they would be stating..'well yes we discovered application detection but it is not affecting quality' instead Hypes it up as if ATI had put the reference to Quake in the driver just for the 8500 launch to inflate benchmarks.

That is the problem...Kyle and his Jerry Springer tatics (although the Spring Break Jerry Springer special is exempt in this case for trash TV)

[Tagrineth]

See, the trouble is, ATi was starting on their unified model, and they probably didn't quite realise that you do have to take different paths for some things, or at least test the paths to make sure both cores behave properly.

Basically the Radeon R6 path that worked perfectly in Q3A, with no IQ problems despite their optimisation on those five textures... had problems with the R200 core; the R200 probably just mis-read what it was being told to render. Remember, soon afterward, they released the next driver which fixed the problem and retained the speed.

[Doomtrooper]

Yep people always forget that, if it was intentional then fixing the IQ and getting better speed then the so called 'cheat' drivers should not be possible.

http://www.hardocp.com/article.html?art=MjU1LDM=

[RussSchultz]

Originally Posted by Doomtrooper

Yep people always forget that, if it was intentional then fixing the IQ and getting better speed then the so called 'cheat' drivers should not be possible.

Maybe, or maybe not. Are we sure it wasn't a case of 'oops, ya got me. I'll try harder (to hide it) next time?'

Or maybe they were working hard on a hardware work around in the driver for something and this was an interim 'patch' to get the the initial benchmarks competitive and later they just found a better way to do it later.

However, I don't think you can simply sweep things under the carpet and say 'oh, it couldn't be intentional because it got fixed later'. All that shows is they found some other way to speed things up--it doesn't disprove that the act was intentional.

And tagrineth, I just don't buy that story. Simply because the word 'quake' was in the previous driver, we cannot conclude that the exact same 'optimization's were present (and presumably working properly) in the R100, yet broke on the R200 and broke in such a specific way that it only affected 5 textures out of the whole bunch.

But, lets explore that a little further: what was the optimization that looked so great on the R100 that didn't work on the R200? What exactly were they doing (beyond "optimizing") that would increase the speed yet not affect the display? Once you've told me that, explain how that didn't work on the R200. I'm coming up blank on something that can answer both those questions.

So far, all I hear is handwaving trying to explain why these 5 textures were problematic and the rest weren't, plus some after the fact rationalizing that it couldn't have been intentional because it got fixed, with a liberal sprinkling of "this isn't as bad as what nvidia is doing".

This is supposed to be the end all be all thread about the truth of the quack/quake matter. Where's those facts?

[Brent]

Originally Posted by Doomtrooper

Yep people always forget that, if it was intentional then fixing the IQ and getting better speed then the so called 'cheat' drivers should not be possible.

http://www.hardocp.com/article.html?art=MjU1LDM=

wow, theres an old review of mine

i remember all that now

notice this part i added back then lol

One thing I do not want to do, though, is get caught up in the whole Quack/Quake issue. I am just going to show you what I have observed first hand that anybody can reproduce and clearly see; just the facts.

[Doomtrooper]

Originally Posted by RussSchultz

Maybe, or maybe not. Are we sure it wasn't a case of 'oops, ya got me. I'll try harder (to hide it) next time?'

Agreed, yet there is no evidence that gives a stronger arguement for either or, yes it could have been a interim patch but the patched driver came out almost two weeks after this fiasco went down.

I didn't ever say ATI was not optimizing, but the Quake Reference was not a specific 8500 optimization...and that is something that was never mentioned by any site.

[Doomtrooper]

Originally Posted by Brent

i remember all that now

notice this part i added back then lol
One thing I do not want to do, though, is get caught up in the whole Quack/Quake issue. I am just going to show you what I have observed first hand that anybody can reproduce and clearly see; just the facts.

One year later your Boss is stirring the pot again though, I don't hold you responsible for some idiotic maneuvers by the 'chief'.

I also don't agree withe [H]'s benchmark methodology

[OpenGL guy]

Originally Posted by RussSchultz

I'm coming up blank on something that can answer both those questions.

That's because you don't have graphics driver experience. It's quite possible that an optimization for one piece of hardware doesn't work on another. Why? Hardware bugs, driver bugs, sometimes a full moon.

I've worked on enough hardware to know that things don't always work as documented (when there is documentation). Then, when you think you've got it figured out, something comes along that upsets the apple cart.

And before you ask what sort of driver bug, picture two closely related pieces of hardware. If you have a bug in your code that misprograms a register sometimes (i.e. some bit fields are the same, but others are different) then you can easily get bizarre results. Been there, seen it.

Again, I have no knowledge of the R200 (aside from what most people here know) or the OpenGL driver, I'm just saying that there are possibilities.

[nelg]

I think that one of the reasons that this issue is so slow to fade from memory is that it has a name, "Quak". The .exe could have been renamed quick.exe or anyting else but someone (perhaps a marketing dep.) came up with a catchy name. I think I will start a new thread looking for suggestions to give Nvidia's new cheat (err. bug) a name. ANy suggestions. Post them here :

http://www.beyond3d.com/forum/viewtopic.php?t=5908

[jjayb]

I was thinking the same thing earlier Nelg. This thing needs a name. Unfortunately I don't have any good names.

3dgate?
Railgate?
Railmark?