[WaltC]

Maybe some of you can help plug the gaps in my recollection of the entire "quack" event, which seems of importance lately only because some folks wish to revive its memory.

What I know of the event, or recall, is as follows:

(1) It became known that a set of ATi's drivers were keyed to recognize the name of the Q3 executable and to implement a less-than-expected rendering quality when Q3 was run.

(2) At publication of this knowledge it was implied that the diminishment of the image quality resulted in greater frame-rate performance in Q3 than was apparent when the Q3 exectuable was renamed and the game run with the appropriate level of IQ.

(3) ATi acknowledged the truth of the driver recognition proposition, but said the diminished IQ was a driver bug relating to specific circumstances in which a bug fix dependent on game recognition for different hardware was improperly invoked.

(4) Ati fixed the bug, and the result was that frame-rate performance was unchanged--that Q3 ran at the same frame rates with the correct level of IQ that it had accomplished with the diminished level of IQ as a result of the original driver bug.

If I've gotten the basics correct here then it would appear to me that there never actually was a "Quack" issue, and certainly not one which is remotely comparable with other situations happening at present. If it is true that ATi did not lose performance when the bug was fixed and IQ restored, then it would appear the entire "Quack" scenario as exposed at the time was at best an error of characterization, and at worst a deliberate fabrication. Any corrections or additions here would be appreciated as I just don't recall the issue that well (or maybe I do.) Thanks.

[PaulS]

That's about my understanding of it, and as far as I'm concerned, the whole Quack thing is a non issue.

nVidiots just like to keep dragging it up as some kind of justification for what nV is/has been doing with their driver "optimisations". See, if your biggest competitior does it, or is seen to be doing it, then it's okay if you completely tear your drivers apart for extra FPS.

Hmm.

[jb]

WaltC

1) those q3 references were in prior sets of drivers before the R8500 release. The "R7500" or original radeons never saw this issue with the messed up textures. Also IIRC it only happened when the texture slider was on its highest setting. And a user reported that it only effected 5 textures total in the whole game.

Weather it was a bug or a cheat we will never really know. One things is for certian that bug should have not made if out of the QA department at the time

[BRiT]

Pretty much on par with how I recall things...

It's really a non-issue.

[WaltC]

Originally Posted by jb

WaltC

1) those q3 references were in prior sets of drivers before the R8500 release. The "R7500" or original radeons never saw this issue with the messed up textures. Also IIRC it only happened when the texture slider was on its highest setting. And a user reported that it only effected 5 textures total in the whole game.

Weather it was a bug or a cheat we will never really know. One things is for certian that bug should have not made if out of the QA department at the time

Thanks much guys....it seems we all remember it about the same way. And thanks,jb, for the added detail--wow, that really brings into focus how much of a non-issue it actually was. Your comments about QA ring a bell, as I still clearly recall my brief, aborted experience with the Rage Fury Pro and its drivers a few years back--stands in sharp contrast to my experience with R3xx and the Cats of the last year. Looking back at that time period I can easily see how this could have been the result of a bug...

[Tagrineth]

We actually had a medium-long discussion about this a while ago (ATi: The Definitive Quake/Quack thread, or something like that, but I can't find it now)...

Basically we came to the effective conclusion that it was originally a legitimate optimisation for Radeon R6 ("R100"), which, in ATi's rushed attempt at releasing the 8500 as fast as possible, got put in the driver (on the assumption and hope that it'd work perfectly) and more or less failed miserably.

[RussSchultz]

Originally Posted by Tagrineth

We actually had a medium-long discussion about this a while ago (ATi: The Definitive Quake/Quack thread, or something like that, but I can't find it now)...

Basically we came to the effective conclusion that it was originally a legitimate optimisation for Radeon R6 ("R100"), which, in ATi's rushed attempt at releasing the 8500 as fast as possible, got put in the driver (on the assumption and hope that it'd work perfectly) and more or less failed miserably.

Actually, no, that wasn't the conclusion.

The essential conclusion was that people asserted lots of things (just like in this thread), but nobody could offer any proof of anything.

Nobody could identify how it was a legimitate optimization gone awry, nor could they identify how the textures that were affected were grouped so that they, and no others were affected. Nobody really could show that the problem didn't exist on the R100. Nobody came up with any information about where those textures were in relation to the maps and the timedemos path.

Essentially, we're all apparently too lazy to do anything ourselves that might shed light on what was going on(myself included) and the people who looked at it when it was current did a piss poor job of really investigating and documenting what was going on.

here's the 4 page thread:

http://www.beyond3d.com/forum/viewto...ighlight=quack

[Ichneumon]

The main reason Quack was such a massive issue for ATI, is that their PR dept just couldn't deal with it with the speed a PR dept needs in these days of 'net "journalism". ATI was giving conflicting responses to those of us who called them up and talked with them about it and a lot of that was because the optimization was *supposed* to be in the driver... however, on the 8500s it caused that bug which arsed up which mip levels were used on certain textures. They just couldn't get the right hand talking to the left hand fast enough to be able to explain that to the 'net before it got way out of hand.

So on one hand ATI was saying it (the optimization/cheat as it was percieved at the time) was supposed to be there, but they weren't clearly saying that the way it was functioning was not the way it was supposed to be.

[H]'s approach to the issue was part of the problem... They didn't contact ATI *AT ALL* before publishing their article, which was hand-fed to them by Nvidia along with the "quackifier" tool for quake3. So ATI was as caught by surprise as everyone else. That doesn't excuse ATI from not being able to be able to react effectively to the issue, which exacerbated the problem.

Thats a damn far change in mode of operation from the [H] we see today, keeping Nvidia issues to themselves for months before coming out and defending them when other sites discover the issues. A true blue (green? ) example yet again of [H] and how beholden they seem to have been to Nvidia for quite a while now.

It was not too long after that that ATI's PR dept was visibly overhauled (change of director and such), and got a *lot* more responsive and up to speed on issues at hand.

Quack was one of the last disasters before ATI's "reorganization" and refocusing their biz with the ArtX aquisition really started showing results.

[Pete]

Originally Posted by Ichneumon

[H]'s approach to the issue was part of the problem... They didn't contact ATI *AT ALL* before publishing their article, which was hand-fed to them by Nvidia along with the "quackifier" tool for quake3. So ATI was as caught by surprise as everyone else. That doesn't excuse ATI from not being able to be able to react effectively to the issue, which exacerbated the problem.

But I remember Kyle just recently saying he was in contact with ATi for weeks/months about the issue, and he published the article to light a fire under their butts--am I wrong?

[jjayb]

But I remember Kyle just recently saying he was in contact with ATi for weeks/months about the issue, and he published the article to light a fire under their butts--am I wrong?

That's what I don't understand Pete. He says he was in talks with ATI about this for months before the article, yet he only got an 8500 days before he wrote the article. How could he have known about this months before the 8500 was out?

[Doomtrooper]

He lies

[Sazar]

Originally Posted by Doomtrooper

He lies

hmm... I believe he was just giving an optimized answer pertaining to the issue when asked about it... question specific optimization I believe...

[martrox]

Revisionist history.......

[martrox]

A couple of thoughts here - IF "Quack" was a cheat (as Russ implys) and not a previous optimization (as Tagrineth and Ichneumon say), then wouldn't it have been a good idea if those that attacked ATI as cheaters offer the proof? That's what is being demanded by those that decry a similar situation now with nVidia. And plenty of proof there is that nVidia does exactly what ATI was accused of, yet I have yet to see anyone disprove the assumption that the real problem with "Quack" was that it was an optimization for the R100 that didn't work properly for the R200. The validity of this is also born out by the timely answer, by ATI, in the form of updated drivers that solved the IQ problem with absolutely no loss in speed.....just try that with ANY of the "problems" found with nVidias' chea....I mean optimizations.

When looked at from this "point of view", how can anyone not see just where Kyle is coming from.

[RussSchultz]

Note, Martrox, if you actually read the thread with an objective eye, rather than the preconceived notion I thought ATI WAS cheating because I disagreed with the assertation that it was an absolute fact that it WAS a bug, you'd notice I was completely open to the idea that it was either--just that there was no absolute facts as some people were asserting.

And I'll also note that its frustrating to see people once again asserting in this thread the "conclusions" of the past, when there were no such conclusions except the ones they had made prior to even having the first discussion on the topic.

[Dave Baumann]

Originally Posted by Pete

But I remember Kyle just recently saying he was in contact with ATi for weeks/months about the issue, and he published the article to light a fire under their butts--am I wrong?

Another Editor tells me that they were offered Quakifier two days before [H]'s article went up. They chose to look at the issue themselves, with their own tools rather than use Quakifier, hence they were later than [H] looking into it. Would Quakifier still be offered around if they knew another site was already looking into the issue?

[martrox]

Originally Posted by RussSchultz

Note, Martrox, if you actually read the thread with an objective eye, rather than the preconceived notion I thought ATI WAS cheating because I disagreed with the assertation that it was an absolute fact that it WAS a bug, you'd notice I was completely open to the idea that it was either--just that there was no absolute facts as some people were asserting.

And I'll also note that its frustrating to see people once again asserting in this thread the "conclusions" of the past, when there were no such conclusions except the ones they had made prior to even having the first discussion on the topic.

Russ, didn't mean to put you on the defensive, we just tend to look at things from totally different directions. Neither is wrong, mind you, just different. While you are frustrated by these conclusions, so am I.... Everywhere I go these days, there are the implications that what nVidia is doing is justified because of Quack - no, it's not said outright, but it is implied. And it does matter - ATI is still taking a beating for this. So, lets talk about what we know.....

1)[H]OCP, for whatever reason, brought up & said, in no uncertian terms, that "Quack" was a cheat. That ATI was using to to get better performace at the cost of IQ.
2)ATI quickly issued new driver that did away with the problems associated with Quack, while still running at the same speed.

So, what can we surmise from this? Was Quack a "cheat", or a mistake? Why use a cheat when you can have exactly the same performance with out the problem? Does this look like ATI had anything to gain by using this cheat? Hmmmm.....

As far as [H]OCP goes, now compare the above with what's been going on with the FX drivers...... I need not list what's going on, but it certianly in NO WAY compares to with "Quack" debacle.

If a website condemns one manufacturer for a small error that was quickly taken care of, while over looking - or worse, maintaining there is no problem - a competitive manufacture of many obvious - found by many different sources - errors that are not only not being taken care of, but compounded by every new driver - well, just what is one to think?

[Dave Baumann]

Does this look like ATI had anything to gain by using this cheat?

Time.

[martrox]

Originally Posted by DaveBaumann

Time.

So, Dave, does this mean you DO think it was a cheat?

[Fred da Roza]

Originally Posted by Pete

But I remember Kyle just recently saying he was in contact with ATi for weeks/months about the issue, and he published the article to light a fire under their butts--am I wrong?

ATI officially announced the 8500 on October 9, 2001.

THG published their 8500 article October 16, 2001. An update article was published November 23 with the driver fixes.

[H] received the card October 16, 2001 and published their article October 19. They published their revised 8500 with the driver fixes December 24, 2001.

I have never personally seen Kyle say he spoke with ATI for weeks/months about this issue before publishing. Do you have a link?

IMO it was important for ATI to get good press out ASAP due to the Christmas time frame. As Dave Baummann suggested, one could argue ATI bought themselves time. Unfortunately for them it bought them a whole lot more.

[Quinn1981]

The 8500 was just a POS when you get right down to it. ATI worked hard to get performance where it needed to be compared to nVidia's offerings. Becuase Quake3 was a common benchmark, it was very important to cheat for them. Later, they worked out performance and didn't have to screw with the filtering. Later, they did the same thing with 3DMark 2001.

nVidia is now doing the same thing. They have flaws with their hardware and try to make up for it by cheating. It's pretty simple and messing with the filtering quality is a easy and effective way to increase performance.

During the 8500 days, I got the feeling ATI didn't care much for their customers, especially after all their big claims they made with the 8500. They had a kind of media black out thing going too, but not nearly as harsh as what nVidia is doing now. ATI fixed their problems to a degree, but nVidia isn't handling it all that well, IMO. I feel they are taking big fat craps on their customers right now just like I saw ATI doing the same.

You have to watch out for yourself with a big company like that, but you should always respect your customers regardless of what you've done wrong. If you could drive video cards, I feel there would be many dead people from side impacts and gas take explosions right now.

[Doomtrooper]

The 8500 was just a POS when you get right down to it

Hardly, a Retail 8500 competes with Nvidias next generation cards like the Ti4200, it plays all current games well, just without FSAA.
The 8500 is superior in so many ways to a Ti card(superior IQ, PS 1.4, Truform, DVD playback), I totally disagree with that statement

[Brandon]

According to ATI, Quack was an optimization gone wrong:

http://firingsquad.gamers.com/hardwa...nt/default.asp

[dan2097]

Hardly, a Retail 8500 competes with Nvidias next generation cards like the Ti4200, it plays all current games well, just without FSAA.
The 8500 is superior in so many ways to a Ti card(superior IQ, PS 1.4, Truform, DVD playback), I totally disagree with that statement

But at the time of quack it was still struggling to compete with the gf3 ti 200 in actual games

[cellarboy]

Originally Posted by dan2097

But at the time of quack it was still struggling to compete with the gf3 ti 200 in actual games

The drivers for the first three months of release sucked hard (especially in XP). However by Jan 2002 they sorted most of the issues out and performance rose masively. From that point on the 8500 was a wonderful card.