[OpenGL guy]

Originally Posted by Unwinder

I bet that you're talking about NV-styled per-stage AF control.

I'll take that bet. How much money ya got? Maybe I'll get my new sports car afterall

[K.I.L.E.R]

Originally Posted by DaveBaumann

I'd be surprised if the hardware could do that.

Why? If the NV20 could why can't the R3xx chips?

[RuslanK]

FUDie

I've seen no evidence that my Radeon 9700 Pro is getting slower in games with these antidetect scripts. Aren't games the reason why we buy these cards?
-FUDie

May be all is ok with 9700Pro, may be not, hope we'll know soon, but do u really think that this change something? Is 9700Pro the only card on the market? The only card from ATI? Seems like both major vendors are detecting benches no matter what PR say and make the cards work with heavily optimized application-specific codepaths. The question is "What do the most of reviewers for several years compare? Card performance or the driver writers cheating abilities?".

Unwinder
73?
nothing could be hidden from you, Alexey 8)

[Dave Baumann]

Originally Posted by K.I.L.E.R

Why? If the NV20 could why can't the R3xx chips?

If you take away the fact that this is being abused to increase benchmark scores, the method of being able to alter the levels of mipmap blending to alter how much "Trilinear" and "Bilinear" is being done between the mipmap levels its quite an elegant one - the idea of getting very close to Bilinear performance, but still removing the most obvious mipmap transitions is quite good. However, as Kristof has pointed out, this quite probably requires hardware support to do it - if ATI could do it then they probably would have by now. This seems to be one of the things that NVIDIA had specifically decided to include in hardware at some point, but its non-stanard and there's not precident for ATI having to include it.

The only problem this is method is that its open to abuse should the IHV(s) that implement it wish to do so, and makes it very difficult to spot when true Trilinear is being utilisied and when its not.

[SirPauly]

I can't comment on the details because it's not set in stone yet, but I will say that things will only get better in terms of quality if the changes are made

I sure hope this comes to fruition.

[demalion]

Originally Posted by Unwinder

First, it's the second of my articles published on Digit-Life (the first was about investigating Intellisample AF optimization techniques in Det 43.45/OGL ). The rest investigation aricles are available on www.nvworld.ru. So there is no need to tell about wrong conclusions typical for DL.

Perhaps this is a failure of communication? Let me specify a bit more clearly: Digit-Life typically provides rather slanted conclusions, poorly connected to the factual support in the article, and your article's conclusion fit that trend (in my observation and evaulation of visiting Digit-Life, and when observing choices and statements not purely factual made there). There was no implication that the conclusion was not yours, just that it was typically as disappointing of the site.
It is quite simply an observation about a lack of logical support provided (clearly specified by the quoted text and description above).

I'm sorry, a solid body of factual support does not give you license to put forth your opinion and agenda (in this case, blaming both companies and being personally offended by ATI) so blithely, and unchallenged, without establishing some coherent correlation between the two. You establish justification for your factual criticism before the conclusion, but none at all that fits the emotional labels and accusations you put forth in the conclusion, and in fact those labels and accusations seem to directly contradict the facts you established (again, I discussed this above).

It's _my_ conclution, I can agrue it and nothing but NV/ATI comments will change it.

It's your opinion, and where you are failing is in not being able to notice the absence of logical support evidenced in the article for it. That makes the label of your opinion as a "conclusion" divorced from the phrase "logical conclusion" that you seem to be assuming is deserved because there was some logic in the article before it.

Unfortunately ATI already had a chance to comment 3DMark2001 issues and they missed it.

And then you passed judgement on them for not prioritizing response enough to suite you. That's an emotional reaction of frustration, and it show through rather strongly in the way your conclusion deviates from the quality of what came before.

I'm afraid that only the ppl who can see "Bad ATI, good NVIDIA" or "Good NVIDIA, bad ATI" may say that the conclusion is wrong.

No, the conclusion is slanted, against ATI if anyone, which directly contradicts the factual support, and which is lent credence in the context of your article only by your decision to focus on 3dmark 2001. You mention the application detection evidence relating to nVidia strongly before, but the greater weight of emotional wording and hostility in the conclusion is allocated in the reverse of the observed facts. You could have slanted the other way ("Bad nVidia, Good ATI", which I assume you meant one of your options to be), but what would have been preferrable would have been to try to either avoid such strong emotional weighting at all, and/or work further to stick to specific context established by the established facts.

Second, the article is really rather emotional. I do feel a lot of frustration after seeng such shame in both drivers.

Yes, that's where you messed up. Read again: you establish clearly more negativity towards ATI than nVidia in your conclusion. That is a rather remarkably skewed approach considering the facts and logic in evidence...you should have recognized that your emotions are not fact and treated them accordingly. Instead, you base stated expectations of behavior for both companies in the future on your emotions instead of the fact, and ended up with a slanted piece. Again, only at the end, IMO...the negative comments about ATI before the end seem to me to be quite reasonable and well supported.

I was not surprised to see a lot of bad things in NV drivers, but it's really sad that ATI benchmark results are also distorted by the driver.

You keep establishing what you did wrong and that you don't see a possibility of a problem with it. You had a greater emotional reaction to your perception of the 3dmark 2001 cheats in ATi's case, because you expected nVidia to disappoint you. That's only because you had reason to expect better from ATI. In point of fact, you did get much better from ATI according to the facts you established, but that is nowhere evident in the emotions that determined your conclusion. What's wrong is that the context of your description of the companies has them starting off at different standards of expectations, which you then apply to the objective facts, and that's just inherently unfair. Even moreso when actually delivering a higher standard of behavior (3dmark 03, lack of extensive application image quality reduction evidence) is the reason you have a higher expectation in the first place. Talk about illogical....you do realize that people with incomplete knowledge will be reading the article too?

OK, I've tried saying that a few times in a few different ways, hopefully one of the times it will succeed in communicating what I'm trying to say. Please realize that this doesn't relate to the opinions you hold yourself (though you might consider looking at things in that perspective, because I think it relates to why you have such friction with OpenGL guy), but with the responsibility of representing those opinions for a reader who doesn't have knowledge as complete as you do, and that your primary goal shouldn't be to lead the reader to think as you do but to give them accurate info and conclusions directly related to facts. The first is done by representing emotional reasons as factual ones and a basis for definitive conclusions at the end of an article....that's what you did, not the latter. Again, a failing typical of digit-life (and other sites, quite often to a worse degree) is in this regard.

I _respected_ their reaction on 3DMark2003 and I really trusted them.

The delta in emotion from trust to what you feel now is represented in the article, but the starting standard of trust is not. You just don't seem to be recognizing that there is a difference yet in your response to me. That change in emotion is not inherently justified such that you can just present it as a factor in your conclusions outweighing the facts you did establish.

As for the CM comments...I read those too, and have the same feeling that it would be disasterous. But did you read the other statements that followed about how that course was stated to be rejected? If so, why, besides your own emotional reaction, were they not represented accurately as well in your expectations for ATI?

I would _really_ like to make different conclusions from such statement and investigation results, but I see nothing but lying in this statement. Nothing more, nothing less.

I'm assuming this is a mistranslation of some sort, because the quoted statement was not a lie, and in fact seemed rather frank and emotional reaction on CM's part. He later indicated that cooler heads and higher ranking personnel had killed the idea dead...you have a right to disbelieve that, but not a right to ignore it or propose an emotional reaction as fact that ignores it.

Are you still trying to assure me that the conclusion is wrong?

This looks, again, like possible miscommunication. I'm saying you were "wrong" to represent your conclusion as you did, for the reasons you did, and with the body of facts you presented, but that doesn't necessarily mean at all that I'm saying your expectation itself is "wrong", though the facts do make it look unlikely at the moment. Your response reads like you are confusing my statements as indicating the expectation itself is wrong (where in fact I encourage, for example, continued inspection of all vendor's drivers for application detections and evaluation of whether they work for or against consumers accurately evaluating the product).

My position is plain and simple: if you're trying to play good guys, do it completely or don't do it at all.

And that's the classic simplification that completely and utterly continues to disappoint me. You make the choices "perfection" or "not giving a damn", and then punish a company more for being closer to "perfection" and ever failing the standard at all, than for "not giving a damn". What kind of behavior do you think results from that type of condemnation?! I'd propose that it is different from sticking to representing thorough factual analysis of what companies did wrong, without focusing so strongly on your own emotional reactions to them, and that the stated purpose of your article isn't served by it either.

I'll continue a long debate of that in PMs if you wish, but I offer you the alternative of exposing what specific cheats the companies did being your focus, and the idea that just maybe letting nVidia get off more easily for "not giving a damn" is not reasonable outside of your own emotional reaction to this issue because you react more strongly against "aiming more towards perfection and falling short".

Finally, it looks like you missed the main idea of this investigation. It's clearly stated in the last paragraph.

Please don't make statements like that, Unwinder. Please take time in reading and trying to process what I said, because being intelligent does not mean that everything you say is automatically evidence of that attribute. I got that the main idea is to call attention to this issue, and before the conclusion you did a good job of that, but then the conclusion further condemned ATI more than nVidia because of your own personal emotional involvement rather than the established facts, and what you ended up doing was serving the end of establishing that competing by more cheating and not admitting or deviating from that course is better than admitting anything or deviating at all from that course. Sorry, that is not a worthwhile point for your article...not because it is ATI you chose, but because the only thing it served fairly was your emotional satisfaction and was a distortion of the facts established in order to do so. Saying I missed something does not do anything to address that.

It has been my observation that plenty of intelligent people are quite capable of refusing to consider (I do not mean "agree with", or "automatically accept", I mean even consider as at all as applicable to what they'd prefer to believe) things said in disagreement to them. Please take a step back, consider the question of what your choice of condemnation in conclusion does differently than your condemnation before it does, and respond to the issue I'm trying to get accross to you rather than with hostility at the idea of criticism. I repeated the same premise at each opportunity where it seemed your commentary failed to consider it, so quote only as necessary to clarify where you don't feel I've established what I object to about your conclusion and why it a valid reason to object.

[Dave Baumann]

You know what, I think if I deleted all of demalion's posts the forum database space usage would probably halve!

[Hanners]

EDIT: Ignore me, I was being an idiot. Worked out what I was doing wrong.

[demalion]

Originally Posted by DaveBaumann

You know what, I think if I deleted all of demalion's posts the forum database space usage would probably halve!

Eventually, the word density will make the database implode anyways. :P

I do try and wait until shorter messages seem to have failed to get my point across.

[digitalwanderer]

To Unwinder:
Did you test the Catalyst 3.5 set for any application detections and is the script compatible with the 3.5 set? (GREAT article BTW, but I haven't gotten a chance to play with the scripts yet....holidays. )

I'm not saying they did, but is it possible that they dropped the cheats from the 3.5 set?

[Pete]

Maybe someone can explain to me why a 64MB 2x2 GF4MX440 is outperforming a 64MB 4x2 9100 by 50% in RtCW at 10x7 plain? I can't believe the difference is solely the GF4MX's superior memory controller.

[Jima13]

Originally Posted by Pete

Maybe someone can explain to me why a 64MB 2x2 GF4MX440 is outperforming a 64MB 4x2 9100 by 50% in RtCW at 10x7 plain? I can't believe the difference is solely the GF4MX's superior memory controller.

and here> http://www.computerbase.de/article.p...da6e53c03#rtcw

you have the 9800pro about 40% faster than the 5900ultra @ 16x12

??

[Pete]

Well, the 9800 has superior fillrate than the 5900 with odd numbers of textures, which RtCW may use. But the 9100 should have about double the fillrate of a GF4MX in *any* situation, so I can't imagine it would lose, much less that badly. Can ATi's OpenGL "implementation" be that inefficient?

[nelg]

Unwinder, I agree with demalion. Perhaps you did misinterpret Catalyst Maker’s post. Look again at the whole post.

OK time to get in the fun.

ITS CM TIME!!!!

First of all as a favor to me, leave Kyle and Brent out of this. This has nothing to do with them. Period

Secondly I am a little upset tonight so I wont say much until tomorrow.

Third I guarantee you that I will ask for an investigation for optimized drivers tomorrow such that has never happened in ATI's history. I am prepared to put a hold on all new features I have in the pipeline so our top engineers can see how much we can optmize by not rendering the whole scene. I am guessing we can gain 25% at this point.
Fourth I am not commiting to do these optimizations ever in a released driver but I think its time for apples to apples comparison.

Fifth I am sorry to hear you (the end consumers) so dissapointed in the state of the industry. I feel for you.

Sixth, you all have my personal guarantee that if you continue to support ATI the way you have so far, I will always be here to help out and be one of the boyz on the forums. (I hope that means something to at least some of you)]

Have a good night everyone and lets talk more tomorrow
Terry

My bold

He admitted to being in an emotional state and there fore was forewarning readers to not take the post literally. This was all in regards to Future Mark renaming cheats to optimizations. Frankly, sarcasm never translates well.

[digitalwanderer]

Originally Posted by nelg

Unwinder, I agree with demalion. Perhaps you did misinterpret Catalyst Maker’s post. Look again at the whole post.

My bold

He admitted to being in an emotional state and there fore was forewarning readers to not take the post literally. This was all in regards to Future Mark renaming cheats to optimizations. Frankly, sarcasm never translates well.

And as a follow up he did later post up to Rage3D apologizing for that post and saying he regretted it very much and that they planned to stop including optimizations in their drivers...or something like that. I'll hunt up the post over on R3D tomorrow if you want, it's late and I'm wiped.

[Exxtreme]

Hi Dave,

Originally Posted by DaveBaumann

I'd be surprised if the hardware could do that.

the Cat3.5 has allready this functionality. I know how to switch between bilinear-af, tri-af with stage optimization and "true" tri-af under D3D.

[Dave Baumann]

How? And can you show us any examples?

[Exxtreme]

Originally Posted by DaveBaumann

How? And can you show us any examples?

There is a registry value named "AnisoType".
You can find it in the registry under:
HKLM\System\ControlSet00x\Services\ati2mtag\Device x when you're using Win2k.

If you set it to:

'0' -> tri-AF without stage optimization
'1' -> bilinear-AF
'2' -> tri-AF with stage optimization.

You need to change your color-depth or reboot your computer when the changes should take effect.

P.S. You can use Demirug's tool to prove it:
http://demirug.bei.t-online.de/D3DAFTester.zip

This tool is similar to Xmas' af-tester but it works under D3D and it allows you to switch between the different stages.

[Unwinder]

Originally Posted by DaveBaumann

If you take away the fact that this is being abused to increase benchmark scores, the method of being able to alter the levels of mipmap blending to alter how much "Trilinear" and "Bilinear" is being done between the mipmap levels its quite an elegant one - the idea of getting very close to Bilinear performance, but still removing the most obvious mipmap transitions is quite good. However, as Kristof has pointed out, this quite probably requires hardware support to do it - if ATI could do it then they probably would have by now. This seems to be one of the things that NVIDIA had specifically decided to include in hardware at some point, but its non-stanard and there's not precident for ATI having to include it.

The only problem this is method is that its open to abuse should the IHV(s) that implement it wish to do so, and makes it very difficult to spot when true Trilinear is being utilisied and when its not.

You've misunderstood me. I ( and probably K.I.L.L.E.R) was talking about the ability to disable trilinear filtering and set the maximum available degree of anisotropy for each textureing stage independently. NV allows to control TLF and the maximum AF level for each texturing stage independently since 28.32. It's ATI turn now.

[RaolinDarksbane]

Unwinder

I have read your article and thought that it was pretty good. I am not tech literate enough to understand everything, but those graphs with description basically explains a lot of thing to me.

The only thing that I have found weird is your conclusion concerning ATI and Nvidia based on those tests. Of all the tests, only one of it was optimized by ATI and it was an old benchmark at that, whereas Nvidia optimized in the majority of the tests.... yet it seems that your opinions on ATI is as bad as Nvidia. That's the only gripe I had with your article, otherwise it was very enlightening.

[Quitch]

As has been said. The article was a good idea, but it suffered from:

a) A lack of range, one software test doesn't tell us a lot.

b) A conclusion that didn't look like to had much to do with the article (and stats) that come before it.

[nelg]

Unwinder, in that thread that you quoted Terry, you also stated that you thought that he was making an emotional response. If you had this suspicion in mind then why did you make a conclusion in which you used his remarks as though they were to be taken laterally

Originally Posted by Unwinder

Terry, I really hope that your reply was rather emotional and you'll change your mind. Don't drop to NVIDIA level, I'm just wondering how both companies are copying dark sides of each other.
It will be the end of Futuremark days, such act will totally discredit 3DMark2003 and it simply won't be used in any reviews (it's exactly what NVIDIA tried to get).

[Unwinder]

Originally Posted by nelg

Unwinder, in that thread that you quoted Terry, you also stated that you thought that he was making an emotional response. If you had this suspicion in mind then why did you make a conclusion in which you used his remarks as though they were to be taken laterally

Correct. At that time I've not seen ATI driver internals and 3DM2001 tricks. Now I see nothing but PR in Terry's 'emotional' statement. Buisness is buisness, it ok to misinform to improve company appearance

[Unwinder]

demalion

I’m afraid that your words are aimed only to rehabilitate one IHV involved in this investigation. You may like the conclusion, you may dislike it, but it will not change utill IHVs will not act differently and your idol will not look better then he currently is. There is only one way to get really fair situation on the benchmarking scene – we need to see the real faces of *all* players to understand the situation. Now we have definitively brash NVIDIA and ATI, which playing ‘good boys’ today. However we’ve already seen similar situation a couple years ago when ATI were trapped on Quack and NVIDIA tried to play ‘good boys’. And now look what’s happened with that ‘good boys’ today? Where are the principles? Do you want the story to continue? I don’t.

Unfortunately ATI already had a chance to comment 3DMark2001 issues and they missed it.
And then you passed judgement on them for not prioritizing response enough to suite you. That's an emotional reaction of frustration, and it show through rather strongly in the way your conclusion deviates from the quality of what came before.

There is no need to judge the things if you don’t know the facts. It’s absolutely not an emotional reaction on them for not prioritizing. It was private discussion dedicated (!!!) to my findings in the driver. And it was direct question about situation with 3DMark2001, that unfortunately remained without answer.

I'm assuming this is a mistranslation of some sort, because the quoted statement was not a lie, and in fact seemed rather frank and emotional reaction on CM's part. He later indicated that cooler heads and higher ranking personnel had killed the idea dead...you have a right to disbelieve that, but not a right to ignore it or propose an emotional reaction as fact that ignores it.

For me the statement ‘Third I guarantee you that I will ask for an investigation for optimized drivers tomorrow such that has never happened in ATI's history’ looks like plain lie. Both competitors have experience in ‘optimizing’ the drivers. Probably you like seeing public lie aimed to improve the company appearance, I don’t.

No, the conclusion is slanted, against ATI if anyone, which directly contradicts the factual support, and which is lent credence in the context of your article only by your decision to focus on 3dmark 2001.

Against ATI? Excuse me but I’ve heard quite the opposite reproofs from NVIDIA fans. The people just see what they want and close the eyes on the rest facts. For me ATI looks against NVIDIA like a nipper against a robber. Furthermore, NV will get a way more problems than ATI from this investigation. But it doesn’t make ATI not guilty and it doesn’t mean that they play honestly now.
However, I really hope that the situation will change one day, ATI have all conditions for fair game because R3x0 don’t need any ‘optimizations’ to compete with NV3x.

Finally, I've made some comments about both IHV's reaction on AntiDetector investigation in this thread:

http://www.rage3d.com/board/showthre...post1332078539

[Dave Baumann]

I’m afraid that your words are aimed only to rehabilitate one IHV involved in this investigation. You may like the conclusion, you may dislike it, but it will not change utill IHVs will not act differently and your idol will not look better then he currently is.

It seems to me that ATI are willing to listen and from the discussions I've had they are actually attempting to move away from as much application detection as possible (even if it a case of legitimately fixing a game, they would rather seek to do that with the developer), however without the buy in from all IHV's then this is just going to meet a dead end.

At the moment ATI has competetive hardware with a relatively clean bill of health (other than 3DMark2001 what else was there?) even in the face of some highly optimised drivers from their competition. At present it seems as though NVIDIA are sitting this one out silently, hoping it will all blow over - so far they've made no commitments that they should remove this types of optimisations, in fact they seem to believe its their right to make them. However, inevitably NVIDIA will get their act together from a hardware perspective and these optimisations will still be there - what are ATI going to do then? Loose sales becuase there may be a reasonable level of hardware parity but because there are no optimisations they look poor in comparison?