AA with NO FRAME HIT on Nv30!!!!!

[gokickrocks]

Originally posted by gstanford
I could not care less what you have to say on DDR-2 compatability modes, unless you can prove you were on the JEDEC board that ratified its specification.
Greg

funny how the person that showed off the DDR2 R300 was the JEDEC president

[Lezmaka]

Originally posted by gstanford
What happens internally with DDR II makes no difference whatsoever what is outside of it - it's like GF3's LMA it does its job of triangle culling whether the app knows about it or not.

Greg

If that were true, shouldn't we be able stick some DDR2 into our desktops with current chipsets? If that were true, then AMD wouldn't need to change the integrated memory controller in Hammer to support DDR2.

[Bigus Dickus]

Originally posted by gstanford
My, we are deep in denial aren't we?

Talk about people who swallow the company line hook line & sinker...

Don't forget: all you have to do is rename quake to quak and all is well again, including your convenient little texture bug.First off, it's Quack not Quak.

Second... of course the texturing bug will dissapear. The code path causing the texturing bug (and performance optimizations) is removed by renaming the references in the driver. You are missing the entire point, or simply avoiding it in order to hold your dear little "cheat" facts close to your heart.

I don't know any clearer way to explain it. All companies optimize drivers. Sometimes that causes bugs, like the QIII texturing bug, or the frequent minor bugs that show up in non-official leaked Detonator drivers. The only difference in the Quake/Quack case is that ATi didn't have time to integrate the optimization into the primary driver path. That was their only "mistake" if you want to call it that. Instead, they programmed an alternate code path (in the drivers, not the game) that utilized the optimizations. These optimizations contained a bug, which was quickly fixed, and the performance remained. The optimizations were later fully integrated into the driver.

Of course, I've already explained this to you several times, but it just isn't sinking in. They didn't cheat, they were just stupid.

[Bigus Dickus]

Originally posted by gstanford
What happens internally with DDR II makes no difference whatsoever what is outside of it - it's like GF3's LMA it does its job of triangle culling whether the app knows about it or not.

Greg What happens internally with DDR-II is what makes it DDR-II. It's the essence of what separates it from DDR-I. You can't change that to "DDR-I compatibility mode." It was designed to look like DDR-I on the outside to facilitate quick integration with existing memory controller designs.

ATi flat out stated that they used DDR-II memory chips on that board. They were running at half the clock and twice the bandwidth per pin internally relative to DDR-I chips. That's what makes them DDR-II. There's no other way DDR-II chips can run internally. ATi demonstrated exactly what they said they did.

Jesus H. Christ.

[The Baron]

OK... analogy time for the Quake/Quack debacle. You're playing basketball, and you know that the other team has one player who A. gets the ball a lot, B. does something really amazing every time, but C. does the same thing every time. Well, how do you defend against that player? Do you treat him like a normal player, or do you adapt your defense to deal with his amazing shot?

If you change your defense, is it cheating? :p

[Megatron]

Originally posted by gstanford
Funny how your "simple bug" only manifests itself on certain ATi hardware using certain ATi drivers isn't it???


Is that kind of like firing up 3Dmark 2001 on a GF4 and watching the scores change depending on whether you enable or disable the splash screens?
:D

Nah...thats probably a "bug"

[StealthHawk]

Originally posted by sancheuz
Once again. The geforceFX wont totally have free AA. But it will probably have the best performance under AA of any card. That is the final answer.

and why is that? because of color compression that ATI already uses? add to that fact that the ATI card has more bandwidth than the nvidia card.

[StealthHawk]

Originally posted by Bigus Dickus
These optimizations contained a bug, which was quickly fixed, and the performance remained. The optimizations were later fully integrated into the driver.

Of course, I've already explained this to you several times, but it just isn't sinking in. They didn't cheat, they were just stupid.

that's really not the only answer. they could have just further optimized the drivers to cover for any loss in performance the new driver revision might have.

we'll never know for sure though. even so, the point is rather moot, why the hell are we having this argument?

Quack had lower performance, probably because some elements of HyperZ II weren't yet implemented fully in the driver. ATI released a new driver, and the performance was the same as with the "cheat" driver. whether it was idiocy or actual cheating who cares if ATI somehow got performance back up to "normal."

[jbirney]

We will never know what happened. But look at all the facts

- Quake3 ref were in the drivers for orignal radeon for 8 months prior to the release of the 8500

- ATI said that they re-used lots of the Radeon driver code

- It was shown that only 5 textures were effected by this "cheat/bug/feature"

- TechReport showed that mip maps were also being fitz with.

- HyperZ2 was not able to be forced on in OpenGL. You could force it on in D3D.

- Renaming the Quake3 exe showed that the IQ improved with the speed feel but up to 20% at high bandwidth settings.


- ATI publicy said there was an bug in the q3 optimizations they did that caused the texutre slider to be read incorrectly

- Two weeks after the launch of the 8500 we had new beta drivers that corrected this "cheat/bug/feature" with realitivly no loss in perfromance and its IQ matched the GF3 serries of cards.

- ATI publicy said that the Q3 optimizations remained but now were applied to all Q3 engines games. Benchmarks on RtCW show this to be accurate as they got a nice boost.


Now if you look at ALL the facts you can not come to the logical conclusion with no doubt that it was a cheat nor can you say with out a doubt it was a bug.

What we have see in the past is fan boys use a sub set of the facts or simple say that one fact does not matter to support his/her own IHV.

The fact that you rename the Q3.exe only proves that what ever happen was happening in the code path that had the Q3 optimizations. That does not prove that it was a cheat nor does it prove it was a bug. Anyone that has every written code will now that it only proves that that part of the code was FUBAR-ed.

Again I don't now what happened as I dont have source code to see what really went on. Neither do the rest of us. So until one of us can read minds I guess we will really never now.

The only fault i have is for ATI not seeing the issue sooner. Something like this is too big to let slip out of the quality department.

BTW we had a nice thread at B3D where a lot of this info came from including ATI employees that worked ont the driver teams. Too bad that was last year when B3D had hosting issues and that thread is long since gone...


I better stop now or Rossco will write another 4000 word essay on the subject :)

[Nv40]

Originally posted by jbirney
We will never know what happened. But look at all the facts


Now if you look at ALL the facts you can not come to the logical conclusion with no doubt that it was a cheat nor can you say with out a doubt it was a bug.

Again I don't now what happened as I dont have source code to see what really went on. Neither do the rest of us. So until one of us can read minds I guess we will really never now.

The only fault i have is for ATI not seeing the issue sooner. Something like this is too big to let slip out of the quality department.

I better stop now or Rossco will write another 4000 word essay on the subject :)

you say JB .... " I don't now what happened "
but at the same time say ATi was not cheating .. :rolleyes:

you are posting nothing more than -> ATI facts..

but that doesnt means this is what really happened .
the problem is that the evidence is so strong that its not possible
to study the quack issue ,and not believe was indeed cheating.

http://www.cs.inf.ethz.ch/37-235/mat/ati-and-ethics.html

QUack thing ,was a way to increase performance without
"noticeable" decrease in image quality .
maybe it was to win time until their drivers were good enough.
but it was clearly a deceptive way to mislead gamers about
a performance that never existed.

lets imagine that the quack issue never happened
or that it was only a "mistake" of a programmer,
that accidentally coded a program that overrided quake3 settings.
what ATi did next again point toward that indeed ATI was
cheating .

the radeon8500 ANiso it just another way
of "cheating" ,which is nothing more that manipulating the
IQ aniso filtering of games to gains performance without "noticeable" decrease in image quality. making the card more atractive for gamers.


http://www.digit-life.com/articles/gf4/index6.html

look at the definition of cheating in the dictionary..
---------------------------------------------------------------------------
1. vt deceive somebody: to deceive or mislead somebody, especially for personal advantage
2. vi break rules to gain advantage: to break the rules in a game, examination, or contest, in an attempt to gain an unfair advantage
-------------------------------------------------------------------------------

just by that definion ATi has been cheating with Nvidia
with their Adaptive aniso method for many years!!
and God know what else . just remember the last benchmarks
i posted.. where the old RAdeon8500 defeated the radeon9700pro in aniso test...

and the sad thing is that this is a "normal" practice of ATI
with their cards . misleading people from the real performance
of their cards . at least this is true now with their aniso test.

what will happen if everyone imitate ATi , and start forcing
adaptive techniques to increase IQ "without noticeable"
decrease in IQ ?

will you complain if SIS XAbre disable 32bit in their games,
"because it not noticeable"for some gamers and runs all games benchamrks in 16bit? defeating ATI and NVidia in all benchamrks?
remember that the diference between 16bit and 32bit for most
people is not noticeable.

Its clear that the RAdeon9700pro is much much faster than the Geforce4 ti4600 ,no doubt . and that card has enough bandwidht
to play with AA+AF all high without problems..

but if the R300 is that faster ,why ATI continues forcing
adaptive algoritms in their aniso ,labeling it as QUality ?
they have all the performance in the world to do it ,the Nvidia
way too.. i tell you why , to make look Nvidia Geforce4 even worse
than they really are. a non adaptive 16x aniso ,using the Nvidia traditional non adaptive way will look much more better ,than the one you already have. and it will take much much more performance too.

yes ,i like performance settings and adaptive methods are good
for getting the best but without "noticeable" ,but in no way that should be forced to us. in no way , just look at rage3d people asking there for non-adaptive methods too. and Hardware reviewers are forced to benchmarks NVidia in 8x aniso quality-fulltime, and ATi with 8x aniso quality-partime.

ATi knows tha most gamers doesnt paid attention to image quality in a scientific way ,but more in numbers in benchamrks.
is always just looking for ways to get unfair advantage over Nvidia in their FOrced ADaptive methods.

So i really hope to see in the future the end of "we know what is better for you " ,labeling Performance settings as quality ,and
allow the gamers the choice of which method they want to use.

but common people lets move to the main TOpic which AA ,
ATi's is not "quacking" anymore in their drivers ,i hope .
and their new Adaptive methods are much better ,than
their old "techniques" . ;)

[Bigus Dickus]

No need for legion here, NV40 has posted the expected fanboy response.

:rolleyes:

[borntosoul]

3d graphics is all about cheating

[Lezmaka]

Originally posted by Nv40
you say JB .... " I don't now what happened "
...i just cut out the irrelevent crap....
;)

Do will you consider nvidia cheating since the GeForce FX has a similar form of adaptive AF?

And give up the whole quack thing. All companies optimize for certain games and game engines. If you think nvidia has never done any optimizations for a game or game engine, then you really should seek some professional help.

[gstanford]

Originally posted by Lezmaka
Do will you consider nvidia cheating since the GeForce FX has a similar form of adaptive AF?

And give up the whole quack thing. All companies optimize for certain games and game engines. If you think nvidia has never done any optimizations for a game or game engine, then you really should seek some professional help.

Do will you consider nvidia cheating since the GeForce FX has a similar form of adaptive AF?
The GeforceFX will has a performance control slider in its drivers (it's already present in the official 41.09 detonators) that allows the user strict control over whether short cuts are taken or not. If you set the slider to "aggressive", it will take shortcuts wherever possible, if you set the slider to "application" everything is done by the book with no short cuts taken.

I've been playing around with the slider the past few days and I doubt I could manually set the driver up to produce better quality than what aggresive settings provide you with - because the driver will only take short cuts where they cannot be noticed.

Greg

[thcdru2k]

huzza? that slider just decides whether the system drivers chooses to use AF / AA or a games ingame settings. for example if you choose no anistropic filtering and used aggressive slider. a game that would allow you to choose whether or not to use anistroic filtering, such as JK II or nba live 2k3, even if af was selected the aggressive settings of the drivers would take precedent and anistropic filtering would be disabled.

[gstanford]

Like I said, I've been playing with the setting for the last few days, and it does deliver more performance on agressive.

I have a GF3 Ti200, and believe me, besides seeing the difference between anisotropic and non anisotropic you can feel the performance difference between modes too.

The driver performance control slider does not override the FSAA and anisotropic sliders beneath it, rather it decides whether to use proper Aniso or optimized aniso, or if a texure even require ansiso filtering, and in the case of FSAA I think it's capable (don't know for sure I stick mainly to quincunx for performance) of dropping to a lower FSAA mode to try to maintain frame rate where needed.

The thing is you can compare screenshots of application controlled setting and aggressive setting and they are all but identical.

Greg

[gstanford]

Keep in mind also the NV30 was designed with adaptive quality settings in mind, so it will make far more use of the settings than GF3/GF4 can.

Greg

[Nv40]

Do will you consider nvidia cheating since the GeForce FX has a similar form of adaptive AF?


-sight-
again , the "adaptive techniques" that video cards companies
does to increase performance at the expense of image quality is not the problem.

Nvidia and ATi and others have performance sliders for that.
and 100% of the PCgames ships with performance sliders too..

so as you see ,there is NO-reason to "optimize" drivers
for particular games. like some companies like to do.

THe problem is when a
company give you NO CHOICE!!! to not use "Adaptive techniques " and at the same time label that setting
the same way as your competition High quality settings ,
just to make their cards faster-> than they really are .
and make others video cards vendors looks bad in benchamrks.
like the example of the RADeon2 vs Geforce4 in aniso test.

i really dont like to talk about this anymore, i posted all this
info because i see again and again people saying it was not
a cheat when the STRONG EVIDENCE say the opposite .
the new RAdeon97xx are much better Hardware
with much less forced "adaptive techniques " ,
and the card has many features thats well worth the money for
the people looking for an upgrade.

[Lezmaka]

Originally posted by Nv40

so as you see ,there is NO-reason to "optimize" drivers
for particular games. like some companies like to do.


Not *some* companies, ALL companies optimize for certain games.

There's no reason NOT to optimize for games if there aren't any adverse effects

[StealthHawk]

Originally posted by Lezmaka
Do will you consider nvidia cheating since the GeForce FX has a similar form of adaptive AF?

only if it drops image quality in certain situations where there is absolutely no reason to ;)