Anand Compares NV38/R360 IQ

[StealthHawk]

http://www.anandtech.com/video/showdoc.html?i=1896&p=1

Basically the conclusion is that Det52.14 are the greatest things since sliced bread and don't degrade IQ at all.

[K.I.L.E.R]

http://www.anandtech.com/video/showdoc.html?i=1896&p=1

Basically the conclusion is that Det52.14 are the greatest things since sliced bread and don't degrade IQ at all.

He's right. Who needs sharp IQ anymore?
QCX AA and bilinear filtering are taking over the industry. ;) :lol:

[3dilettante]

I don't really see any serious image faults in the det 52.14 screenshots, but I can't really see anything with the picture sizes and how they are cropped. The links to the full-sized pictures don't seem to work for me, and not all pictures had that option.

In the case of Aquamark, the grass did seem pretty fuzzy though. There was some evidence of inferior AA, but it was pretty hard to see with the samples they showed.

It is also unfortunate that a number of the programs being used don't seem to be able to capture the same frame with each sample.

It's possible the det 52.14 drivers don't significantly degrade the image quality, but I guess it will take more time to find out if they really hold up as they should after they are released.

[Laa-Yosh]

Forget it, I was wrong ;)

[Laa-Yosh]

Finished reading... well, Anand certainly had a busy week.

Unfortunately many screenshots are cropped and/or taken with a camera angle so that you really can't take a look at various IQ related things (like texture filtering, AF). There aren't any obvious differences between the ATI cards and the NV cards using det 5x.x drivers. Also, neither the Halo nor the TR images had anything that looked like using PS2.0 to me.

I don't know what to think about the NV38 being unreleased though... or about using an AthlonFX for the tests.
And then there's the question of where Nvidia has managed to gain performance from, or how Halflife 2 would perform. Also, I know this will sound paranoid, but I truly, honestly hope that other websites will be able to reproduce Anand's benchmark results.

[mczak]

IMHO this article isn't that bad. Though I'm now wondering why anand even did part1, as this article has all benchmarks mentioned there included.
And the article failed to mention some of the "minor" issues, I'd really have liked to have some analysis of the tri/af (quality/app pref) behaviour of the different drivers.

And I don't agree with (some of) his conclusion remarks, well of course if nvidia can get a 30% increase in performance with every driver at some point will they beat the r360! But looking at the hardware, this just doesn't seem possible...

The X2 benchmarks seem also pretty useless (anand mentions it himself), so why put them up in the first place? Especially if the problems encountered are likely to change the performance numbers (and just dropping frames as anand suggested sure would change the numbers).

Anand also somehow manages to blame ATI for the missing shiny water in NWN, since he was obviously browsing around some forums he should have known that bioware just used nvidia proprietary extensions and didn't bother to also code for ATI's extensions. (For the record, shiny water in nwn still doesn't work with linux and ati cards (and ati's drivers). Some bioware person said in the nwn forum the ati shiny water code isn't in the linux build, since nobody at bioware bothered to test it - maybe it would just work...)

Will there be some analysis of this driver on beyond3d (these should be whql'd sometime iirc)? I'm sure anand missed some things. 8)

[Natoma]

Hmmm.... Given the mention of image anomalies in almost every test in part 1, yet his screenshots show almost nothing at all, could this be the result of that screen-cap detect stuff Valve complained about?

I wonder.....

[Brandon]

Heh, I found this part from page 2 to be ironic: "Our previous Flight Simulator benchmark just didn't push the game far enough, and we are hard at work trying to find a benchmark that better reflects gameplay and is completely repeatable".

Someone really needs to tell Anand that using benchmarks that rely on game cutscenes (Halo's are actually letterbox cutscenes!) isn't reflective of real gameplay.

[Laa-Yosh]

Given the mention of image anomalies in almost every test in part 1,

Wow, I've completely forgot about that...

yet his screenshots show almost nothing at all, could this be the result of that screen-cap detect stuff Valve complained about?

Would be truly shocking... but he has looked at the tests (at least he says so), so he should've noticed it...

[Natoma]

Laa-Yosh,

I put nothing past anand at this point. Imo the burden of proof is on him to do a thorough analysis.

At the very least, no mention whatsoever of that very damning point from Valve, in an image quality test of those very drivers, strikes me as "curious" at best.

[Dave Baumann]

:shock: Fuck me, that was a lot!

Generally I thought it pretty good.

They didn't want to delve into any test apps, or even use mip colours, so they didn't spot the trilinear issues. I thought it odd that they didn't wait an just use 3.8 straight away (don't ask how this could be done...). I'd also like to know why they made the assumption that NVIDIA would be the only ones to gain from a shader optimiser - ATI are only beginning this as well.

[Laa-Yosh]

Here's what Anand wrote in Part 1 of the article:
The 52.14 drivers apparently have issues in two games, neither of which are featured in our test suite (Half Life 2 & Gunmetal).

Gunmetal screenshots in Part 2 do have some anomalies, actually on all the three, there are some artifacts...

Moving on with Part1:
AA and AF didn't really seem to work as well on the NVIDIA cards as it did on the ATI cards. There was some difference between the two, but we will have to do more research into this area before we can bring forth anything conclusive.

No mention of it in Part 2.

Homeworld2:
This is another test where it was not apparent that NVIDIA's AA was doing as much as needed for the scene, so we will take a closer look at this benchmark when we do our image quality comparison.

Now he simply does not include AA scores in Part 2.


So it looks like he has covered most of the IQ issues mentioned in the first part of the article. Sim City 4 is an interesting question though...

[OpenGL guy]

I found his treatment of TRAOD to be in rather poor taste. For all the reader knows, none of the cards could give a playable game because he only shows how much performance drops when enabling PS 2.0. I mean, are we going from 100 to 40 or 10 to 4? Also, you can't really compare the results between the 9800XT and the 5950 because you have no idea where they started without PS 2.0.

[Laa-Yosh]

Proper examination of the AM3 screenshots show the NV one to be much blurrier. Save and view, or use your browser's back/forward buttons:

http://images.anandtech.com/reviews/video/roundups/2003/Fall/Part2/aquamark/nv52.14.allon.jpg

http://images.anandtech.com/reviews/video/roundups/2003/Fall/Part2/aquamark/cat3.7.allon.jpg

[Natoma]

Now that's what bothers me. Those two pictures are supposed to be with AA enabled, and yet the 52.14 drivers don't look like AA is being applied at all to me. No mention in the blurb above the images.

And the blurriness is noticeably marked between the two images in the textures...

[Fred da Roza]

IMHO this article isn't that bad. Though I'm now wondering why anand even did part1, as this article has all benchmarks mentioned there included.
And the article failed to mention some of the "minor" issues, I'd really have liked to have some analysis of the tri/af (quality/app pref) behaviour of the different drivers.

~

The X2 benchmarks seem also pretty useless (anand mentions it himself), so why put them up in the first place? Especially if the problems encountered are likely to change the performance numbers (and just dropping frames as anand suggested sure would change the numbers).


I thought this was going to be an IQ comparison. It seemed to be mostly benchmarks with a few comments about IQ here and there.

I know Anand mentioned the lack of AA in Homeworld and X2, but it looks like it's not working in F1 either. Just look at the car. And what's with the tiny images.

http://www.anandtech.com/video/showdoc.html?i=1896&p=12

[OpenGL guy]

Now that's what bothers me. Those two pictures are supposed to be with AA enabled, and yet the 52.14 drivers don't look like AA is being applied at all to me. No mention in the blurb above the images.
There's AA in the 52.14 shots, it's just very poor quality (Lack of gamma correction and ordered grid).

[Pete]

:shock: Fuck me, that was a lot!

Generally I thought it pretty good.

They didn't want to delve into any test apps, or even use mip colours, so they didn't spot the trilinear issues. I thought it odd that they didn't wait an just use 3.8 straight away (don't ask how this could be done...). I'd also like to know why they made the assumption that NVIDIA would be the only ones to gain from a shader optimiser - ATI are only beginning this as well.
I thought the R3x0 architecture has almost-optimal drivers as-is? That's what I've gathered from these forums, anyway. I recall a recent thread where ATi devs said not to expect 20%+ performance jumps in shader-heavy games simply due to optimized drivers.

Yes, I agree that Derek (who was the main author) didn't spend enough time remarking on general AA/AF quality, and I didn't like his screenshot treatment. I noted this in his Ars thread, so hopefully he'll give me his reasons for that. I hope they're nothing more than time pressure, but I still prefer quality over timeliness.

[Natoma]

Now that's what bothers me. Those two pictures are supposed to be with AA enabled, and yet the 52.14 drivers don't look like AA is being applied at all to me. No mention in the blurb above the images.
There's AA in the 52.14 shots, it's just very poor quality (Lack of gamma correction and ordered grid).

Hmm.. Definitely poor quality then. I guess I'm spoiled by my 9800 Pro. :lol:

Interesting comment from Derek Wilson in the comments section for AT:

Many of the image quality issues from part 1 were due to rendering problems that couldn't be captured in a screen shot (like jerkiness in X2 and F1), or a lack of AA. For some of the tests, we just didn't do AA performance benchmarks if one driver or the other didn't do what it was supposed to.

That sniffs very strongly of the screen-cap probs Valve reported. IQ problems in game that can't be detected in screenshots.

[Natoma]

:shock: Fuck me, that was a lot!

Generally I thought it pretty good.

They didn't want to delve into any test apps, or even use mip colours, so they didn't spot the trilinear issues. I thought it odd that they didn't wait an just use 3.8 straight away (don't ask how this could be done...). I'd also like to know why they made the assumption that NVIDIA would be the only ones to gain from a shader optimiser - ATI are only beginning this as well.
I thought the R3x0 architecture has almost-optimal drivers as-is? That's what I've gathered from these forums, anyway. I recall a recent thread where ATi devs said not to expect 20%+ performance jumps in shader-heavy games simply due to optimized drivers.

Yes, I agree that Derek (who was the main author) didn't spend enough time remarking on general AA/AF quality, and I didn't like his screenshot treatment. I noted this in his Ars thread, so hopefully he'll give me his reasons for that. I hope they're nothing more than time pressure, but I still prefer quality over timeliness.

Sireric noted in another thread that they've only begun scratching the surface of the shader compiler. Basically shader rendering to this point is brute force of the hardware and un-optimized.

[Joe DeFuria]

I found his treatment of TRAOD to be in rather poor taste. For all the reader knows, none of the cards could give a playable game because he only shows how much performance drops when enabling PS 2.0. I mean, are we going from 100 to 40 or 10 to 4? Also, you can't really compare the results between the 9800XT and the 5950 because you have no idea where they started without PS 2.0.

That is my single biggest gripe with this review. WTF was up with the TR performance difference without the absolute numbers? :?:

And while most of the gross rendering issues with the Dets (on these tests) appear to be fixed, there is no image quality analysis on filtering, which is a huge question mark.

I certainly don't expect a scrutinizing of each of the tests...but at least pick one or two and go into some detail. (Particularly one or two tetsts that had large performance increases.)

A lot of time and effort went into that, and it's appreciated...but there was hardly any "image quality comparisons" to speak of.

Not even a lick bout relative AA quality (when nVidia was doing AA, that is.)

The overall tone seemed to be almost apologetic to nVidia in many cases...though I'll re-read the article with a more open mind before commenting on some of it. ;)

I mean, based on the following:

1) ATI lead the majority of tests, and some by a significant amount
2) ATI had far fewer "driver issues" encountered

ATI is the clear winner....and while he states his preference for ATI, it comes with the "you should really wait" disclaimer. "Wait till next year" to decide the DX9 performance winner?

How about decide the performance winner now....and if it happens to change next year, so be it?

[Dave Baumann]

I thought the R3x0 architecture has almost-optimal drivers as-is? That's what I've gathered from these forums, anyway. I recall a recent thread where ATi devs said not to expect 20%+ performance jumps in shader-heavy games simply due to optimized drivers.

I think its more of a case of "don't expect it yet". AFAIK, 3.7's were the first release of any kinds of optimiser, however there are some gains to be had yet - if you think about the drop in GT4 in 3DMark03, going from hand coded shaders to unoptimised, then there is still that type of gain to be had. Eventually the automatic optimiser should get close to the performance of the hand coded shader. They still have some work to do to be able to automatically reorder code to make best use of co-issue, better use of the mini and full ALU's and hiding texture ops within ALU ops (as these can happen in parallel with R300).

[Laa-Yosh]

I suggest everyone to save the images and go through them one by one. F1 challenge AF is noticeably better on the Radeon for example.
The naming conventions used for the images have to be changed though because they create a mixed-up order...

[Natoma]

I found his treatment of TRAOD to be in rather poor taste. For all the reader knows, none of the cards could give a playable game because he only shows how much performance drops when enabling PS 2.0. I mean, are we going from 100 to 40 or 10 to 4? Also, you can't really compare the results between the 9800XT and the 5950 because you have no idea where they started without PS 2.0.

That is my single biggest gripe with this review. WTF was up with the TR performance difference without the absolute numbers? :?:

C'mon now Joe I pegged you as smarter than that. Must you ask this question? :wink:

And while most of the gross rendering issues with the Dets (on these tests) appear to be fixed, there is no image quality analysis on filtering, which is a huge question mark.

I really wanted to hear about the screen-cap problems Valve brought up to see whether or not those apply.

ATI is the clear winner....and while he states his preference for ATI, it comes with the "you should really wait" disclaimer. "Wait till next year" to decide the DX9 performance winner?

Because TR:AOD is not a "true" DX9 game and HL2 has been delayed conveniently, maybe, into next year.

*cue conspiracy music

Dunh Dunh Dunnnnnnhhhhhhhhh

How about decide the performance winner now....and if it happens to change next year, so be it?

You expect far too much my child.......

[Natoma]

I suggest everyone to save the images and go through them one by one. F1 challenge AF is noticeably better on the Radeon for example.
The naming conventions used for the images have to be changed though because they create a mixed-up order...

Indeed. What happened to the nice rollover images anand used to use in his articles?

[Laa-Yosh]

Copy-paste of my post from anandtech:

Note: the AA/AF and noAA/AF images of Warcraft3 have been mixed up for the NV52.14.

It tells a lot about the value of the screenshots that it takes careful inspection to find this error. I have played a lot of War3 recently and the difference is very noticeable in game, even with this GF4.

[RussSchultz]

That sniffs very strongly of the screen-cap probs Valve reported. IQ problems in game that can't be detected in screenshots.

I really wanted to hear about the screen-cap problems Valve brought up to see whether or not those apply.

no mention whatsoever of that very damning point from Valve, in an image quality test of those very drivers, strikes me as "curious" at best.

could this be the result of that screen-cap detect stuff Valve complained about?


You wouldn't be obsessing would you? (And these are just from this thread!)

Many of the image quality issues from part 1 were due to rendering problems that couldn't be captured in a screen shot (like jerkiness in X2 and F1), or a lack of AA.

OMG! NVIDIA is hiding jerkiness by purposely capturing a still image! OMG! ;)

[Natoma]

I am obsessing about that when it comes to an IQ comparison chock-full-o-screenshots. Those problems were enough for Valve to publicly state to everyone "Do not use these drivers!"

You're not interested in the least bit about that issue?

p.s. I've read the comments on the AT comments board. I did not make the comment about Valve, surprise surprise. The past two reviews, someone has taken my comments and posted them verbatim to their forum, and some AP also insulted my intelligence thinking that I was posting. :lol:

Didn't realize I was that special. :shock:

[Joe DeFuria]

http://www.anandtech.com/talkarticle.html?i=1896&ATVAR_START=41&p=3

The post at the top of that thread (not by me...I did not post in that thread at all), pretty much sums up my feelings on the "apologetic tone" toward nVidia with examples.

Now back to some more productive and pointed observations:

Aquamark, AA/Aniso image quality.

Clearly, there is some "blurring" going on in the nVidia shots..including alpha textures (grass). There's only two possible explanations for this:

1) The AA mode used by the nVidia card is one of the XS / partial supersampling modes.

2) The AA mode used by nvidia is Quincunx.

Since the whole image appears blurred to me, not just "filtered" relative to the ATI shot (I mean every texture if you compare them), I'm inclined to believe it's Quincunx.

Is there more than one Quincunx / bluir filter AA mode on NV cards? Or just the 2X AA + blur?

[THe_KELRaTH]

I presume the question whether the FX59xx is using AA or not in some images is due to only having Vertical offset AA. Therefore areas like the Michelin / lights bar in F1, the X2 animated logo will look as though AA isn't on - and circular graphics such as the Stop sign in Simcity etc won't look smooth as only half the circle has AA applied. - (More like HSAA)!
I suppose it's another NV performance saver when comparing competitor AA tests.

[Sxotty]

You know I hate to say it, but I am rather impressed, I thought the FX line was milked completely out pretty much, getting gains with that little drop in IQ, I think is actually an achievement of note. However it still doesn't make them competitive and nothing short of a new arcitecture will do that though, because even though (despite what many claim) nvidia is still decent for most current games they cost to darn much.

[zurich]

I'm beginning to think that alot of the hardware community's hatred towards NVIDIA has become so blinding, they will never appreciate anything that comes from the company now, or in the future. Not that I think that NVIDIA hasn't made mistakes, but I think alot of people have taken things so far that even if the companies were reversed a year from now, nothing could change the ATI love fest.

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

It's just pretty messed up. 52.xx shows large improvements. Anandtech praises these impovements (while still giving the final nod to ATI), yet gets accused of being on the NVIDIA payroll. Pretty scary times these days.

This isn't so much an obervation on Anand's article, but rather on the responses I've read around the net regarding it.

*insert joke about people taking cheating/lying to heart so seriously yet overlooking the current American administration, meh (or something)*

(except Natoma, that crazy DemocrATIc :lol:)

[Ailuros]

Joe,

There's also 4xOGMS + 9-tap; by the way the bluriness of Quincunx or the 4/ 9-tap mode really never came across in all it's glory in screenshots. The latter mode is just a tad more apparent on screenshots, but it's rather mild to what you experience in real-time. If you have really time to waste, re-read both the first and the second article; the oxymoron is quite obvious between them.

Personally I'll start objecting to any review from now on that won't use custom timedemos. Unless I've missed something I can't seem to see any at Anand's.

Wavey,

Wouldn't it be safe to assume though that the performance gain with shader optimizers is quite a bit higher with NV3x's than with R3xx's?

[Ailuros]

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

If you really want me to look at it objectively, I'll give you my personal opinion about the specific driver set when I can download it myself and see it's behaviour with my own eyes.

The NVIDIA "hatred" you're addressing there is an ongoing phenomenon for years now, nothing new there either.

[jjayb]

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.


You've got a short memory then. I remember a time when every ATI card review had the disclaimer that their drivers weren't as good as Nvidia's. Also the disclaimers about "Nvidia has something better just around the corner". Nothing has changed here except the roles have been reversed.

[BRiT]

I'd also like to know why they made the assumption that NVIDIA would be the only ones to gain from a shader optimiser - ATI are only beginning this as well.

It's simple. They're genuine twits.

[GraphixViolence]

I'll have to add myself to the list of skeptics about the supposed improvements offered by the Det52.14 driver that Anand tested. If this driver is really so fantastic, why isn't it available for download? Maybe because it still has too many issues? Maybe because it isn't passing the WHQL certification tests because it's messing with IQ and/or DirectX instructions? Maybe because it doesn't yet provide enough performance to match ATI's latest products, even with Nvidia's own unreleased future hardware?

I think the last question is particularly interesting. How many people would actually spend $500 on a graphics card that's not the fastest AND doesn't have the best IQ? Would it even make sense for Nvidia to announce the NV38 and Det50 until they've found a way to somehow get higher performance than the 9800 XT in a significant number of games? Perhaps convincing gullible press like Anand to spend countless hours reviewing of unannounced hardware and software is their new "paper launch" strategy... this way, you can start generating hype about your upcoming products without having to actually commit to anything like clock speeds, availability, features, etc.

[silhouette]

ATI is the clear winner....and while he states his preference for ATI, it comes with the "you should really wait" disclaimer. "Wait till next year" to decide the DX9 performance winner?

How about decide the performance winner now....and if it happens to change next year, so be it?

:lol: I remember the exact quote somewhere else.... (hmm.. was it in the Anand's article comment section? )

[nelg]

Sixty pages of Sizzle. Unfortunately no Steak. I seemed to get the impression that so long as the IQ. was close enough it is fair to compare. Personally I would rather see ten games done throughly than twenty glossed over.

[Bjorn]

I'd also like to know why they made the assumption that NVIDIA would be the only ones to gain from a shader optimiser - ATI are only beginning this as well.
Not that i agree with it, but Ati's hardware has been around a lot longer so i guess they're assuming that they have less things left to optimize for.
But i don't really understand this part though:

If NVIDIA can continue to extract the kinds of performance gains from unoptimized DX9 code as they have done with the 52.14 drivers (without sacrificing image quality), they will be well on their way to taking the performance crown back from ATI by the time NV40 and R400 drop

Nvidia's hardware runs at a 50 MHz clock advantage (9800 XT vs 5950) and also uses >25 million transistors more and they're still not able to beat Ati, with a lot crappier FSAA to boot. So i don't know how the Det 52.xx drivers can lead anyone to believe that they are on their way to take back the performance crown. I would say that there's now a possibility that they can get it back though (again, assuming that these performance gains are legit).


How about decide the performance winner now....and if it happens to change next year, so be it?

He does say:
NVIDIA has a long road ahead of them in order to improve their compilers to the point where game developers won't have to hand-code special NV3x codepaths, but for now ATI seems to have won the battle.

And the conclusion seems ok to me:


ATI is still the recommendation, but NVIDIA is not a bad card to have by any stretch of the imagination. We still urge our readers not to buy a card until the game they want to play shows up on the street. For those of you who need a card now, we'll be doing a value card round up as part of this series as well.


I'm very interested in the value card round up since i want to know if these new enhancements will carry over to the entire FX line. Hopefully we'll soon see some Det 52.xx reviews that includes 9600, 5200-5600 series. Perhaps from B3D ?

[3dilettante]

Is AT going to put the cards in the lower tier into a third article?

Hopefully they will include some IQ comparisons for those cards as well, as it would help show if there are any optimizations being made solely to boost the top end's benchmark scores. The graphs in the latest article seemed kind of sparse without the cards that were featured in the first part.

[mboeller]

According to the new 3DCenter.de article :

the new Deto 52.10/52.14 is worse than the Deto 51.75

when it comes to image quality; especially when using AF.


So I'm a little wondering how Anandtech came to his conclusions.

Link : http://www.3dcenter.org/artikel/detonator_52.14/


sorry for being so loud.

[3dilettante]

Is the pyramid demo in the 3dcenter article a custom demo? If so, there is a marked lack of improvement versus the standard timedemos.

Nice to see some screenshots that don't require me to squint.

[nggalai]

Is the pyramid demo in the 3dcenter article a custom demo? If so, there is a marked lack of improvement versus the standard timedemos.

Nice to see some screenshots that don't require me to squint.Yes, it's a custom timedemo, i.e. a recorded MP match.

Sorry that there's only a German version. Should there be any troubles understanding the article (the images speak for themselves, mostly), feel free to ask questions here:

http://www.beyond3d.com/forum/viewtopic.php?p=177096#177096

Or open a new thread here (no registration required):

http://forum-3dcenter.de/vbulletin/forumdisplay.php?s=&forumid=77

93,
-Sascha.rb

[3dilettante]

Is the pyramid demo that much more demanding on the system than the antalus flyby?

The performance figures seem pretty steady between the scores of the pyramid levels, but the antalus flyby has a dip in the no AA/AF portion of the 52.14 dets. I wonder what causes that dip.

In a side note: The full size screenshot links are working in the AT article. It's just too bad only a select few programs have that option. What's up with the lack of Tron screenshots? Perhaps IQ isn't a concern according to them, but I like screenies.

[nggalai]

Is the pyramid demo that much more demanding on the system than the antalus flyby?Much, much more demanding. It's like a botmatch run with no botmatch AI, but with all the models, shadows, explosions, gunfire lighting etc. Have a go at it yourself:

http://www.3dcenter.de/downloads/ut2003-pyramid2003.php

There's still some CPU usage due to the physics engine, but basically, the Pyramid demo is mainly GPU bound.

93,
-Sascha.rb

[whql]

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

Look at this objectively - ATI won 90% of the benchmarks in there, and some by a not inconsiderable margin, and 90% of the conclusion was apologising about NVIDIA's performance, even with the suggestion that they may be faster at some point in the future. I'm sure the single line "ATI is still our recommendation" is of so much joy to the ATI hardware and software engineers who've spent their time making hardware that beats NVIDIA in 90% of the games tested (and without IQ degration). :roll:

[cthellis42]

Quite lengthy, but I was rather hoping for a broader range of Dets and Cats to see play and perhaps some refrast examples. And a LOT more full-size image testing.

Sadly the whole matter is MUCH more than a pair of reviewers can do, so we're going to have to wait on 52.14 to run the gaming gamut and pull specifics from ALL the sites out there. Thankfully there are a lot of amazingly anal observers out there, so given a few weeks and a LOT of reading we'll have the best idea. ;)

[Hanners]

OMG! NVIDIA is hiding jerkiness by purposely capturing a still image! OMG! ;)

:lol: :lol:

That is too funny... If that isn't a sig-worthy quote, I don't know what is! :wink:

[Doomtrooper]

According to the new 3DCenter.de article :

the new Deto 52.10/52.14 is worse than the Deto 51.75

when it comes to image quality; especially when using AF.


So I'm a little wondering how Anandtech came to his conclusions.

Link : http://www.3dcenter.org/artikel/detonator_52.14/


sorry for being so loud.

Eh ?? Well that is simple..3Dcenter know what their talking about :lol:

[demalion]

...

I mean, look at this objectively;

OK

the 52.xx performance gains are massive.

Without regard to gains being relative to achieving the same output, I do think this is a fair objective assessment.
But it seems to me that if this is done without achieving the same level of output, such "performance gains" is best characterized as "fictional". Therefore, it seems important to me to take a reasonable amount of steps to answer that question. I think AT failed.

The IQ quality compared to 4x.xx is again a huge improvement.

Hmm...I don't quite get how you propose this statement as factual, such that you make it a fundamental premise of an "objective" evaluation.

I have a problem with this statement being proposed as such due to an abundance(!) of factors:

A concentration on the Anti-"Synthetic Benchmark" mantra, while completely ignoring the "fixed camera path problem" that is the real issue nowadays, and which is more likely to be exposed with dedicated ("Synthetic", nevermind how representative the workload) benchmarks. I feel like someone is trying to sell me a bridge, and some reviewers seem to have bought it already.
Selective presence of full screen images for reader controlled comparison. It takes work to reduce images after capture, so it isn't like they didn't have them to begin with. It also seems to me that you can "cut and paste" any "click to view original" code, so it wouldn't be more work to include them. This leaves me with the impression of selective truth to facilitate Anand's conclusion in place of the reader being given full information to come to their own, which in my experience is one effective tool of "marketing" and/or "lying". The rollover issue someone else mentioned also concerns me, as well as the idea of jpeg based comparison of filtering and AA in still screen shots...especially with so many being only available as reduced to 450x450 (as far as I could determine). :-?
[list:49e5e70d81] Aquamark 3 : NO full size image.
C&C Generals : full size images.
F1 Challenge : NO full size images (with an AA/AF comparison, even!).
Gunmetal : NO full size images.
Halo: full size images (NO AA/AF comparison...no evaluation, illustration, mention, or consideration of shader effects at all, which you think might be warranted with the large amount of confidence Anand speaks invests in the 65% performance gains. Ah, but even though they can relate to graphics workload, cut scenes don't represent gameplay anyways...but do the screenshots? What about representing compiler technology or card performance, as Anand uses them for (especially in his conclusions about compilere technology)...which relates to the fixed camera path issue I'll cover in more detail later.
Homeworld 2 : NO full size images (with an AF comparison...AA didn't work for the nVidia cards, though resampling to 450x450 seems likely to make that a bit irrelevant for the screenshots).
JK- Jedi Academy : full size images (and finally with an AA/AF comparison. But hold on, they recommend you set them in game, but then went and set them in drivers, though in game settings are exactly where a difference should be most evident between ATI and nVidia...if it weren't a jpeg still screen shot and there were more than one texture layer used in the game (?). It is the only full size image AA/AF comparison that it seems possible for the reader to evaluate effectively, yet their selection of how to test it doesn't seem logical or consistent with even their own statements, nor with issues surrounding nVidia and ATI in general.
NWN : NO full size images (with another AA/AF comparison).
SimCity 4 : NO full size images (with another AA/AF comparison).
TR-AOD: The most competent and thorough explanation in the review, AAICS, and full size screenshots offered for examining PS 2.0 output. What disturbs me is the lack of an actual thorough PS 2.0 performance comparison between ATI and nVidia (from any benchmark, it wouldn't have to have been TR-AOD) to go with what they use TR-AOD to show. Least logical comment in the discussion : "I know everyone will want to take these numbers and say that they universally describe the DX9 performance hit on NVIDIA hardware, but we have had plenty of other benchmarks today that show very different results." In contradiction to his discussion of how specific to TR-AOD his analysis is, it is even used to validate every other shader benchmark in a way inconsistent with the data shown.
UT2k3: NO full size images (and Yet Another AA/AF comparison at the reduced size). No controversy around this game and filtering...let's just show the full size images for JK instead. :shock: Did I just miss the full size images being offered for this? That would make more sense than what I think is the case now.
WC3: NO full size images (with an AA/AF comparison).
Wolfenstein-ET: NO full size images (with an AA/AF comparison).
X2 : Not sure what the screenshots really mean. Looks like something in the balancing act I've thought AT has tried before, where (IMO) the "alternate" IHV is shown in the best possible light in one targeted instance to shield from accusations of bias, even when that requires doing something that doesn't even seem to make sense. ATI's AA is better, but what does showing us a screenshot of a high contrast logo demonstrating this really tell us, when you propose it represents a game? Why not a screenshot from the models/graphics for the rolling demo?
OK...maybe I'm looking at old pages, or a complete collection of full sized images were offered elsewhere and I missed it. Perhaps someone could correct me if such is the case, or I made some errors in the above. As it stands, I'm just a bit boggle by the inconsistencies, and that still leaves the following...
This next list is made especially poignant In the context of the above, and conclusions like this: "The very large performance gains we saw in Halo, Aquamark3, X2 and Tomb Raider can be attributed to the enhancements of NVIDIAs compiler technology in the 52.xx series of drivers", when this relationship is made up on the spot in the conclusion.
A lack of answer to the some of the specific questions (borrowed from Valve's presentation, and related outside of that context as appropriate) seems a gross oversight:
Camera path-specific occlusion culling.
Independtly confirmed to have been done by nVidia before. On the camera path, image quality is (usually) identical, but completely unrepresentative for performance evaluation or comparison. Big problem when this image quality comparison doesn't even seem to take the rudimentary step of custom timedemos to begin to try and avoid this happening. :?:
Visual quality tradeoffs (lowered filtering quality, disabling fog).
Hmm...I'm pretty sure I've seen "4x.xx" drivers without filtering issues, though the fog question is still unanswered for me. I think Anand could have answered it, but I don't recall him trying. This could, however, be something that the new drivers improved.
Screen-grab specific image rendering.
This remains unconfirmed outside of Valve's statements, AFAIK. Does seem to throw another monkey wrench in limiting the selection of AA/AF comparisons that users can effectively evaluate for themselves, while at the same time proposing they are universal indicators of "improved image quality" and skipping consideration of other AA/AF issues in your conclusion. As if there weren't enough wrenches in that already.
Lower rendering precision.
Big problem, where was the investigation of this? Perhaps they meant B3D to be used for this, but the comparison they linked to doesn't seem to answer that question. Nor do they seem to consider it for their conclusions.
Algorithmic detection and replacement.
Actually, with the HLSL compiler SINCOS issue, I think there is a valid case for nVidia to do this at the moment. Unfortunately, I don't think nVidia would limit themselves to validly addressing such problem when they did do shader replacements. Of course, if you use Cg and don't get performance gain compared to HLSL, this should remove that concern, AFAIK. In any case, see the above issue with where I think AT went wrong with evaluating the concerns this brings into it.
Scene-specific handling of z writes.
I'd say this is a big problem in reality, but not a big problem for AnandTech failing to check this (at the moment, with the lack of tools we have, or at least that I know about).
Benchmark-specific drivers that never ship.
I find certain displayed "logical conduct" regarding driver usage policy at AT some reason for major concern. However, this is only a big problem if the drivers used by Anand are actually never released, despite "assurances". This could end up being moot, however, if they do so...except, of course, for the "HL 2 performance update" blog entry (which, however, wasn't actually in an article, though the conclusion to this one reminds me of the same mindset).
App-specific and version specific optimizations that are very fragile.
I think this is more of a "big problem" for a game that is going to be updated very frequently, when performance will vary greatly as these optimizations "break", unless the developer agrees to freeze their shaders for the convenience of an IHV making custom replacements :-?. I know this is/was a big problem for HL2, but I don't know how much of an issue it is for other games...except as far as having to wait for nVidia to get around to your game, and making patches a possible nightmare for developers. :-? Also, the uncertainty as to whether nVidia will be able to have success in creating an equivalent shader (outside of those games and timedemos used as popular benchmarks) or not.[/list:u:49e5e70d81]

I don't really think much of this article, overall...quality over quantity. There seems to have been a good opportunity and frame work for some really good work here, but with spotty follow-through, though a lot of time seems to have been spent. The use of reduced size images seems ridiculous to me (I'm still wondering if I just missed something?), and even if that wasn't an issue, there is a complete failure to addresse even the simplest of the last list of concerns, as well as a significant lack of coherency in the "logic" present in the conclusion.
Where there is full sized IQ analysis, the choices of methodology and performance comparison presentation seem to be reasonably characterized as being chosen selectively to prevent showing nVidia in a bad light, and then applied as a universal basis for evaluating all other information, even where the same things were specifically not evaluated in what it speaks about (AA/AF and the relative performance of the "PS 2.0 compiler" between nVidia and ATI).
The (absent) performance comparison (not the same thing as "performance hit relative to itself comparison") between ATI and nVidia in TR-AOD would seem an excellent opportunity to test the idea of nVidia's ability to "take the crown" from ATI as they propose in their conclusion, yet its not even clear to me if this comparison is without shaders completely, or with minimum settings. Not that I could tell anything from the screenshots anyways (way too dark), but I'm assuming it is due to my monitor that I'm replacing.

This, to me, is large scale deception and/or significant incompetence. Perversely, the strength of what they did show with TR-AOD (I don't have an issue with it, I have an issue with what it is used in place of and the information that ends up being witheld due to that) seems to argue against the one of those I personally find the least disappointing.

Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

Where do you get the idea that nVidia is "hated" because they are behind? Maybe some diehard nVidia fans "hate" them for that (?), but I think the hostility towards nVidia I'm most familiar with is for reasons besides that.

It's just pretty messed up. 52.xx shows large improvements. Anandtech praises these impovements (while still giving the final nod to ATI), yet gets accused of being on the NVIDIA payroll.

The issue isn't as simple as "the reported fps gains are greater" and "Anand says the image quality is improved".

Pretty scary times these days.

That conclusion does not seem even remotely objective to me, nor does it seem to have been determined based on objective evaluation.

To be fair:

This isn't so much an obervation on Anand's article, but rather on the responses I've read around the net regarding it.

I know there are some people who go to extremes in hostility, but they happen to have been given a rather large amount of cause. For instance: I didn't generally find it "scary" that people condemned ATI vehemently for "Quack", and there was quite a bit less behind the issues going on then. I did, and do, find some of the "logic" surrounding both issues a bit "deficient", "lacking", and therefore "surprising" to me in that way...but that has to do with (some of) the reasoning, not the hostility itself.

My concerns above relate to both matters.

[Laa-Yosh]

First of all, the small images are not resized/resampled, only cropped. You can see aliasing artifacts on them well enough... assuming that there are any.


Homeworld 2 : NO full size images (with an AF comparison...

Also, there aren't any surfaces that would show any serious texture aliasing, because of the camera angle. Useless for AF comparison.

JK- Jedi Academy : full size images (and finally with an AA/AF comparison.

However, the lack of contrast and detailed patterns in the floor/wall textures basicaly hides any texture aliasing / bluring. Useless for AF comparison.

NWN : NO full size images (with another AA/AF comparison).

See Homeworld 2.

UT2k3: NO full size images (and Yet Another AA/AF comparison at the reduced size).

Once again, there are practically no surfaces in the view that would show any texture aliasing. You can still see that something isn't right with the NV5x drivers, but there's not enough information to make any observations.

WC3: NO full size images (with an AA/AF comparison).

See Homeworld2. The camera angle wouldn't show any aliasing... also note that they've originally mixed up the NV5x images, as I've previously noted. And if you actually play the game, the texture aliasing is very disturbing, especially in cutscenes.

Wolfenstein-ET: NO full size images (with an AA/AF comparison).

See UT2003.


Conclusion: the majority of the screenshots are simply useless for any image quality comparison. Some of the OK images show noticeable problems with the NV5x drivers.
The more I look at this article, the more disappointing it becomes. I seriously hope that B3D and ExtremeTech will get their hands on the new drivers + a GFFX ASAP.

[demalion]

First of all, the small images are not resized/resampled, only cropped. You can see aliasing artifacts on them well enough... assuming that there are any.

Phew. I'm glad this was me missing this...resampling would have been a bit "remarkable". Picking portions of the screen while deciding to selectively refuse to offer the original screen images is quite problematic enough. I guess squinting would have paid off after all, at least as far as evaluating what was actually shown, and outside of any resolution disparities.

[digitalwanderer]

According to the new 3DCenter.de article :

the new Deto 52.10/52.14 is worse than the Deto 51.75

when it comes to image quality; especially when using AF.


So I'm a little wondering how Anandtech came to his conclusions.

Link : http://www.3dcenter.org/artikel/detonator_52.14/


sorry for being so loud.

Eh ?? Well that is simple..3Dcenter know what their talking about :lol:
ROFLMFAO~~~~

Best line I've read this morning, thanks for the laugh...could you please explain that to Evan Lieb next time you see him? I'm having the damnest time trying to beat some sense into that fellow. :roll:

[THe_KELRaTH]

A nice guy at AT forum, hominid skull, has put an animated Gif of the F1 images. http://www.aclift.eclipse.co.uk/example2.gif

I'm suprised AT didn't do this as it's alot easier to view any differences - even with those tiny images.
AT's conclusion should have included that if you don't want line crawling even with AA on then of the reviewed cards ATI is your ONLY option right now.

[Mintmaster]

A nice guy at AT forum, hominid skull, has put an animated Gif of the F1 images. http://www.aclift.eclipse.co.uk/example2.gif

I'm suprised AT didn't do this as it's alot easier to view any differences - even with those tiny images.
AT's conclusion should have included that if you don't want line crawling even with AA on then of the reviewed cards ATI is your ONLY option right now.

Wow, I thought it was just AA that was ATI's big strength in F1, but looking at the white road line, you can see that NVidia compromised AF as well. I guess this confirms the results of the 3Dcentre article. Doesn't it look like NVidia is only using 2x horizontal AA? If it was 4x (even ordered grid), shouldn't you at least be able to see one medium blue color on horizontal surfaces, like the Michelin sign?

Anyway, I always found that racing games are where you need AF the most, and it can even improve your racing ability if the road lines disappear without it.

[digitalwanderer]

I'm suprised AT didn't do this as it's alot easier to view any differences - even with those tiny images.
I don't think Anand's goal was really to show the differences, I think he was trying to hide them...thus the tiny screenshots. ;)

[[3dc]Leonidas]

Is the pyramid demo that much more demanding on the system than the antalus flyby?Much, much more demanding. It's like a botmatch run with no botmatch AI, but with all the models, shadows, explosions, gunfire lighting etc. Have a go at it yourself:

http://www.3dcenter.de/downloads/ut2003-pyramid2003.php

There's still some CPU usage due to the physics engine, but basically, the Pyramid demo is mainly GPU bound.

93,
-Sascha.rb


Pyramid2003 benchmarks with 8xAF + 4xAA:
http://www.3dcenter.org/artikel/r9500+9700+9800_vs_gffx5200+5600+5800/zj_bench_ut2003_d.php

Without AA and AF the Pyramid2003 timedemo ist CPU limited on highend graphic cards. But with AA and AF it is highly gfx limited (7,5 fps on GFF5200). Anyway, this demo wasnt taken to show a CPU or a gfx limitation, it was taken to show UT2003 real gameplay performance.

UT2003 Timedemo with more gfx limitation:
http://www.3dcenter.org/downloads/ut2003-overkill.php

[Quitch]

Was it just me, or did a couple of the side objects look a little sharper under the old dets over the cats? Couldn't say, since I'm not sure what they're supposed to look like :)

[Sazar]

Was it just me, or did a couple of the side objects look a little sharper under the old dets over the cats? Couldn't say, since I'm not sure what they're supposed to look like :)

the dets show much more blurriness than the cat 3.7 if you look into the distance.. there are also very apparent 'boundaries' where filtering is being applied it seems..

[Dio]

A nice guy at AT forum, hominid skull, has put an animated Gif of the F1 images.

AT's conclusion should have included that if you don't want line crawling even with AA on then of the reviewed cards ATI is your ONLY option right now.
I can't see any AA at all on two of those three images.

[andypski]

Looking at the original Anandtech 4xAA JPG shots from F1 Challenge (.gifs really aren't good for image quality comparisons, and JPGs suck as well) either he did something wrong or there's something strange going on as there appears to be no aa on the horizontal edge of the banner over the track at all on either the 45.23 or 52.14 shots. I would expect to see one intermediate shade even on nVidia's 2x mode (it should stretch about halfway across the stairstep between one y coordinate and the next), and in the 4x mode there should be mostly one intermediate shade stretching about the same distance with the odd pixel of other shades here and there, but there seem to be pretty much no intermediate shades at all.

There are a few pixels dotted around some of the other edges in the image that look like they do have some level of AA applied though.

Very curious. :?

[THe_KELRaTH]

If you grab the F1 FX images and zoom in there is one intermediate shade on the banner aswell as other horizontal offset areas (like the car front window), areas that are near the vertical axis have 4 shades.

EDIT here's a closeup - I think I'm just seeing an odd pixel
www.jlmay.f2s.com/F1%20Zoom.JPG
and here's a reminder pic of NV30 v R300 AA
www.jlmay.f2s.com/NV30-ATIAA.jpg

NV Horizontal offset AA..... Where?

[Bouncing Zabaglione Bros.]

I'm suprised AT didn't do this as it's alot easier to view any differences - even with those tiny images.
I don't think Anand's goal was really to show the differences, I think he was trying to hide them...thus the tiny screenshots. ;)

Taken with the in-game view carefully angled to minimise the problem. I guess it's a case of "Nvidia tells us it doesn't impact image quality" :roll:

[Headstone]

the one thing that stood out for me was on the 52.14s the last fence panel on the right. If you look at the 3.7s and the 43s they both show the trees through the fence. On the 52s you get grey.

[andypski]

If you grab the F1 FX images and zoom in there is one intermediate shade on the banner aswell as other horizontal offset areas (like the car front window), areas that are near the vertical axis have 4 shades.
Not on the horizontal edges - as far as I can see there are a couple of pixels (at most) at each stairstep on the banner that might be an intermediate colour - in either 2x or 4x AA modes there should be a long strip of an intermediate shade running about halfway between each stairstep and the next (because of the sampling patterns used). There might be a similar problem with the horizontal edges of the tires as well, but because of the underlying noise from the road surface it's difficult to tell.

It's very strange.

Looking at it some more there also doesn't seem to be much detail in the road surface in the 52.14 drivers AF shot (looking at the area of track nearest to the car the track 'noise' looks to be at a much lower frequency)

[Edit - the lower frequency on the track detail appears to be new on the 52.14 driver - the 45.23 screenshot doesn't have this effect.]

[Natoma]

Hey andy while you're here, quick question. :)

Does ATI lap the R3x00 core before applying it to the circuitboard? I noticed that the core on my 9800 pro was unbelievably reflective when I took off the stock heatsink to apply arctic ceramique; almost a perfect mirror. Never seen that before on any core, processor or VGA alike.

[THe_KELRaTH]

Oh dear... Looks like CatalystMaker's gonna piss off AT. All that designer TR-AOD benchmarking's just gone out the window with Cat 3.8 :lol:

" Tomb Raider frame rates improve as much as 20%"

[RussSchultz]

Hey andy while you're here, quick question. :)

Does ATI lap the R3x00 core before applying it to the circuitboard? I noticed that the core on my 9800 pro was unbelievably reflective when I took off the stock heatsink to apply arctic ceramique; almost a perfect mirror. Never seen that before on any core, processor or VGA alike.
The back of a die/wafer is generally that shiny. At least all the wafers I've seen.

[andypski]

Does ATI lap the R3x00 core before applying it to the circuitboard? I noticed that the core on my 9800 pro was unbelievably reflective when I took off the stock heatsink to apply arctic ceramique; almost a perfect mirror. Never seen that before on any core, processor or VGA alike.

I don't know so I can't tell you I'm afraid - you could try sireric.

[Quitch]

the dets show much more blurriness than the cat 3.7 if you look into the distance.. there are also very apparent 'boundaries' where filtering is being applied it seems..

I agree that in the centre of the image, the road and the overhead signs looked much better in the cats... but I couldn't really tell with the side buildings which was giving the better image there (since without knowing what the building SHOULD look like). Only thing I can say is the new dets were pretty (and obviously) bad.

Looking at it like that though, the cat is the obvious winner.

[Socos]

I'm beginning to think that alot of the hardware community's hatred towards NVIDIA has become so blinding, they will never appreciate anything that comes from the company now, or in the future. Not that I think that NVIDIA hasn't made mistakes, but I think alot of people have taken things so far that even if the companies were reversed a year from now, nothing could change the ATI love fest.

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

It's just pretty messed up. 52.xx shows large improvements. Anandtech praises these impovements (while still giving the final nod to ATI), yet gets accused of being on the NVIDIA payroll. Pretty scary times these days.

This isn't so much an obervation on Anand's article, but rather on the responses I've read around the net regarding it.

*insert joke about people taking cheating/lying to heart so seriously yet overlooking the current American administration, meh (or something)*

(except Natoma, that crazy DemocrATIc :lol:)

Just from the AA aspect of it I don't think Nivida is competing at all. Why even turn it on for a FX card? I'm not blinded by anything. If anything I see more clearly than ever.

I guess the questions you should ask yourself is this: 1. Does AA quality matter to me even on the horiziontal? 2. Do I want my card to perform the best it can out of the box, or wait six months to a year to see good performance?

If the answers to both of those questions are yes then you have your answer.

If you answer NO to both questions then you my fellow are the one that is not seeing clearly. :roll:

[Hanners]

I'm beginning to think that alot of the hardware community's hatred towards NVIDIA has become so blinding, they will never appreciate anything that comes from the company now, or in the future. Not that I think that NVIDIA hasn't made mistakes, but I think alot of people have taken things so far that even if the companies were reversed a year from now, nothing could change the ATI love fest.

I mean, look at this objectively; the 52.xx performance gains are massive. The IQ quality compared to 4x.xx is again a huge improvement. Yes, ATI is still ahead slightly, but that never caused anyone to 'hate'/witch hunt them when they were behind in the DX8 days.

It's just pretty messed up. 52.xx shows large improvements. Anandtech praises these impovements (while still giving the final nod to ATI), yet gets accused of being on the NVIDIA payroll. Pretty scary times these days.

This isn't so much an obervation on Anand's article, but rather on the responses I've read around the net regarding it.

*insert joke about people taking cheating/lying to heart so seriously yet overlooking the current American administration, meh (or something)*

(except Natoma, that crazy DemocrATIc :lol:)

I think what you have struck on here could be regarded as the heart of nVidia's problems now - After about twelve months of deception, cheats, and image quality reductions, people expect that and nothing more from them, and because of that are naturally wary and skeptical of anything that comes out of nVidia. nVidia need to start earning back the respect they had in the past - The Detonator 50s might be the beginning of that long road, they might not. Just remember how long it took ATi to regain trust and respect after their reputation for poor quality drivers and the whole Quack incident.

Personally, I think there will be a lot of genuine improvement in the Detonator 50s, but that is going to be tempered by more image quality reductions in other areas - From what I've read it's going to be very much a mixed bag.

[[3dc]Leonidas]

English translation is ready:
http://www.3dcenter.org/artikel/detonator_52.14/index_e.php


Please keep in mind, ATi have a similary optimization under Direct3D AF: texture stages 1-7 are only bilinear filtered. IMO this is not corrected with Cata 3.8.


2006: We have GeForceFX III / Radeon 20000 and play Doom 7 with a nice bilinear filter at all :-)

[OpenGL guy]

Leonidas]English translation is ready:
http://www.3dcenter.org/artikel/detonator_52.14/index_e.php


Please keep in mind, ATi have a similary optimization under Direct3D AF: texture stages 1-7 are only bilinear filtered. IMO this is not corrected with Cata 3.8.
This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.
2006: We have GeForceFX III / Radeon 20000 and play Doom 7 with a nice bilinear filter at all :-)
It's funnier if you get your facts straight.

[K.I.L.E.R]

That reminds me GL Guy, those ideas you told us about a CP option to force trilinear on all texture stages; have they taken off yet? :D

[THe_KELRaTH]

Is it reasonable to assume that the jump in DX9 shader performance is due to some form of wrapper so calls for Full precision auto drop to partial and calls for partial auto drop to 16Int.

[Xmas]

This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.
That's a bad excuse IMO. The number of games that take control of the AF themselves is quite small. Luckily aTuner (http://www.3dcenter.org/atuner/index_e.php) can disable that "optimization". But something isn't automatically good if something else is worse.

At least OpenGL has been exempted from any filtering "optimizations" so far.

It's funnier if you get your facts straight.
Maybe his comment was badly worded, but I don't think Leonidas has to get his facts straight.

[Xmas]

Is it reasonable to assume that the jump in DX9 shader performance is due to some form of wrapper so calls for Full precision auto drop to partial and calls for partial auto drop to 16Int.
No, that would be easy to spot. If the driver drops to lower precision at all, it needs to do a more thorough analysis of how the values are used to decide where a drop in precision would be "acceptable".

I don't think performance increases of 30% with complex shaders in comparison to early drivers are particularly surprising. Many shaders leave room for optimization, especially those written in HLSL.

btw, it's FX12, not 16.

[OpenGL guy]

This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.
That's a bad excuse IMO. The number of games that take control of the AF themselves is quite small.
And the number of games that benefit from trilinear on every stage is small.
But something isn't automatically good if something else is worse.
I have no idea what this means.
It's funnier if you get your facts straight.
Maybe his comment was badly worded, but I don't think Leonidas has to get his facts straight.
No, I believe he thought that ATI never did full trilinear, which is not the case. Many people have gotten confused on this issue.

[[3dc]Leonidas]

This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.


Jap. But, how much apps have own AF settings? 20%?

[BRiT]

Jap. But, how much apps have own AF settings? 20%?

Well in particular, the one major game you and everyone else seems to be testing with, UT2K3. So why you state the blatant flasehood that ATI does the same optimizations as Nvidia when the App controls AF, I have no idea.

[StealthHawk]

This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.
That's a bad excuse IMO. The number of games that take control of the AF themselves is quite small.
And the number of games that benefit from trilinear on every stage is small.

Are there any games that benefit from not doing trilinear on all stages? From what I've noticed, apps that gain performance also seem to lose IQ(Aquanox2, Max Payne, UT2003). If there are games that benefit from not doing trilinear on all stages without an IQ decrease, can you name some specific games?

[Bambers]

from AT shots nvidia appears to have problems using 4xAA (or capturing it) in more than just homeworld2, F1 is also missing AA.

for that matter he said that there were no noticable differences in jk:ja where the cat3.7s are missing the sabre glow :?

[Zvekan]

Leonidas]
This is only true if you've forced AF from the control panel. If you use application enabled AF, then you get what the application asks for.


Jap. But, how much apps have own AF settings? 20%?

I think you are missing the point. I don't remember reading anywhere that enabling AF via control panel offers full trilinear on all stages, so it is ATI's decision to make filtering per stages as they like.

But it is important to say that if application specifically asks for trilinear to be carried on all stages and you enable application AF you will get what applications asks for.

It is not ATI's fault that gamedevelopers don't offer AF sliders in their game - write to developers of your favourite game and ask them to add additional functionalty in upcoming patches.

Zvekan

[Xmas]

That's a bad excuse IMO. The number of games that take control of the AF themselves is quite small.
And the number of games that benefit from trilinear on every stage is small.
I can make this decision for myself, thank you.

Maybe his comment was badly worded, but I don't think Leonidas has to get his facts straight.
No, I believe he thought that ATI never did full trilinear, which is not the case. Many people have gotten confused on this issue.
Read the last paragraph of this article (http://www.3dcenter.org/artikel/ati_nvidia_treiberoptimierungen/index9_e.php) written by Leonidas.

[Xmas]

I think you are missing the point. I don't remember reading anywhere that enabling AF via control panel offers full trilinear on all stages, so it is ATI's decision to make filtering per stages as they like.

While I can understand your point (It's a driver override, so you can't expect the application settings to work), I don't agree with it. The AF slider should control AF, not mipmap interpolation.
And besides that, it's not about whether this "optimization" is valid regarding the semantics of the control panel, it's about the user not being able to get the best possible quality out of the hardware, and about knowing it when trying do do a fair comparison.

[Dave Baumann]

That's a bad excuse IMO. The number of games that take control of the AF themselves is quite small.
And the number of games that benefit from trilinear on every stage is small.
I can make this decision for myself, thank you.

Control panel filtering is a privelidge, not a right. Technically speaking its outside of WHQL certification to provide these at all, since it should be entirely up to the app to specify what is selected.

[madshi]

Control panel filtering is a privelidge, not a right. Technically speaking its outside of WHQL certification to provide these at all, since it should be entirely up to the app to specify what is selected.
Maybe, but I thought ATI wants to be the IQ king. If you look at all the reviews, most people are not even aware of ATI's "optimization" (not too long ago you even had to tell Brent about it, if I remember right). And the people that are aware usually don't like it. I find it disappointing that the IQ king doesn't offer the highest possible IQ which the hardware can do.

[Dave Baumann]

But they do - if the application requests it.

[madshi]

But they do - if the application requests it.
Yes. But only then.

Dave, wouldn't you like to have the possibility to force full IQ via the control panel? You sound like you wouldn't care!

[Xmas]

Control panel filtering is a privelidge, not a right. Technically speaking its outside of WHQL certification to provide these at all, since it should be entirely up to the app to specify what is selected.
You're certainly right, but I'm not a lawyer, I want to play games with the best quality possible. And if I get 100+ fps with mipmap banding, I'm certainly going to complain about it - which I have a right to, btw :wink:

[Dave Baumann]

Yes. But only then.

Dave, wouldn't you like to have the possibility to force full IQ via the control panel? You sound like you wouldn't care!

Sadly, this is the effect, not the cause. I'd like it if every one of the IHV's offered quality options that didn't make me, as a reviewer, chase my tail every time a new set of drivers are sent out. Lets be frank - we all know why this is there and we all know why it isn't going to change anytime soon.

[Bouncing Zabaglione Bros.]

Yes. But only then.

Dave, wouldn't you like to have the possibility to force full IQ via the control panel? You sound like you wouldn't care!

Sadly, this is the effect, not the cause. I'd like it if every one of the IHV's offered quality options that didn't make me, as a reviewer, chase my tail every time a new set of drivers are sent out. Lets be frank - we all know why this is there and we all know why it isn't going to change anytime soon.

Are you hinting at the fact that IHVs like to control the options so that they can ensure their cards arn't made to look bad? Or are you saying that developers are just too lazy to provide proper controls?

What I can't understand why we are still seeing so many games arrive with very limited in-game options to control things like AA/AF & filtering levels. Are the developers not interested in having people set their game visuals properly, rather than through a global setting that may or may not make their games look like trash?

[Bambers]

Afaik the reg entry anisotype can still be used to force full trilinear in all, while it would be nice to have all 3 settings in the drivers it would only really annoy me if it became impossible to force all trilinear from the registry.

[tEd]

Are you hinting at the fact that IHVs like to control the options so that they can ensure their cards arn't made to look bad? Or are you saying that developers are just too lazy to provide proper controls?



ut2003 flyby is to blame

[[3dc]Leonidas]

Control panel filtering is a privelidge, not a right.


Yes and no. Its NV´s privilege to do, whatever they want. And its my privilege, to buy theirs gfx - or not. And if NV/ATI dont offer a full trilinear + anisotropic filter on 80% of all apps ...


PS: Dont ask the programmer of a 1999 game for an anisotropic filter. Ask the programmer of the 2003 gfx driver.

[[3dc]Leonidas]

But they do - if the application requests it.
Yes. But only then.

Dave, wouldn't you like to have the possibility to force full IQ via the control panel? You sound like you wouldn't care!



Ack. I dont want force AF in 100 apps. I want one CP with one setting for all.

[WaltC]

Control panel filtering is a privelidge, not a right. Technically speaking its outside of WHQL certification to provide these at all, since it should be entirely up to the app to specify what is selected.
You're certainly right, but I'm not a lawyer, I want to play games with the best quality possible. And if I get 100+ fps with mipmap banding, I'm certainly going to complain about it - which I have a right to, btw :wink:

You don't have to be a lawyer (thankfully) to understand the control panel is a universal, generic approach to IQ settings which the IHVs have implemented to address the shortcomings of 3d games published without a user interface to control in-game settings. There's a simple reason it can't be perfect running all 3d games without such internal settings controls: all 3d game engines are not equal and don't do things the same way. So when they set up a Cpanel they put in place the best settings for a majority of 3d games without internal settings controls. That is simply the best it's possible to do for Cpanel global IQ settings controls.

If the game developers were doing their jobs we wouldn't even need a Cpanel IQ-settings interface. But they're not, so that's why the IHV's include them, and have for a long time. I will say, though, that if anything the IHVs are guilty of doing too good a job with the Cpanels, because even software developers become overly dependent on them instead of building user settings controls into their games as they should be doing. The developers who think those settings are properly the province of the IHV are wrong, IMO.

The problem with UT2K3 and the Cat Cpanel settings is a case in point. When the Cpanel is set to Application Preference the game can properly instruct the Cats as to which texture stages need to be trilineared, and everything works as it should. When the Cpanel is invoked it overrides the game engine commands to the driver and the global setting is used, and only one texture stage is trilineared. Obviously, then, UT2K3 is an object lesson on why it is better to let each game control the driver's IQ settings through internal settings controls, since you are then dealing with specific commands relative to specific game engines instead of generic Cpanel configurations which are configured based on IHV "best estimates" of what's needed in the majority of 3d games without such controls.

(Obviously, this is an entirely different situation from the current nVidia control panels, which are geared to ignore certain stage trilinear filtering commands whether the user asks for them though the Cpanel or through such in-game settings control as might apply. Much different situation there.)

Another thing about UT2K3 and user IQ settings within the game. Their implementation in the software from the standpoint of a general user is very poor, and documentation from Epic on user adjustment and the conventions and parameters used in thier .ini text files is non-existent, AFAIK. At least, such useful documentation as might ship with the games. That, however, has nothing to do with any IHV. That's Epic's failing, IMO. But to give them credit, Epic at least includes internal user adjustable settings for its engine, which many 3d games simply ignore.

My feeling is that IHVs like ATi should stress to developers that their Cpanel settings are generic, and that if they wish for the Cats to support things outside of the limited general scope of the Cpanel in their games, they need to build in the necessary in-game settings controls so that their customers can get the kind of IQ the developer ostensibly wants his product to display. Some people have suggested that IHVs start including things like Cpanel controls for selecting which texture stages are treated in a game, but I think the IHVs have already made it too easy for game developers to ignore the inclusion of complete in-game IQ feature support. With complete in-game IQ feature support, these things aren't needed in the Cpanel, and game developers need whatever incentive can be provided for supporting API IQ settings in their software products. IMO, of course..:)

[Dave Baumann]

Leonidas]Yes and no. Its NV´s privilege to do, whatever they want.

My earlier statement made no mention of anyone doing whatever they want in the control panel - on the contray in fact. The very fact that it is there shows they are enabling a greater level of IQ control at beyond MS's certification practices. What is not acceptible is not providing the full level of filtering if/when an application selects it.

PS: Dont ask the programmer of a 1999 game for an anisotropic filter.

I wasn't. But, how many games from 1999 and before actually display issues with the filtering levels available through the control panels?

[vb]

Leonidas]But they do - if the application requests it.
Yes. But only then.

Dave, wouldn't you like to have the possibility to force full IQ via the control panel? You sound like you wouldn't care!



Ack. I dont want force AF in 100 apps. I want one CP with one setting for all.

not trying to show off here (not english native) but that would not be forcing AF in 100 apps, that would be requesting it.

and I would rather have that on a low/mid end card than playing with options in the CP and trying to remember WTF worked decent for every god damned game i decide to try.

and, yes, the CP thing is forcing. ever tried to play Splinter Cell with a certain CP setting?

but i agree that we should have aniso+tri as an option even if it is buried "very" deep.

[[3dc]Leonidas]

But, how many games from 1999 and before actually display issues with the filtering levels available through the control panels?


Zero. But we dont talk about IQ issues, we talk about max IQ.


The problem with the current ATi/NV drivers: You cannot play 80% of games with a full trilinear/anisotropic filter. A 500-$-gfx cant do a full trilinear/anisotropic filter in 80% of games. Mmh, incorrect: The hardware can do this, but ATI/NV doesnt want you to use this potential.

[BRiT]

[3dc]Leonidas, Absolutely not true. With the ATI drivers and a registry setting, you can force max AF for all texture stages.

[StealthHawk]

Leonidas]Zero. But we dont talk about IQ issues, we talk about max IQ.


The problem with the current ATi/NV drivers: You cannot play 80% of games with a full trilinear/anisotropic filter. A 500-$-gfx cant do a full trilinear/anisotropic filter in 80% of games. Mmh, incorrect: The hardware can do this, but ATI/NV doesnt want you to use this potential.

Iff IQ is not increased by doing trilinear on all texture stages, then why does it matter for those 79-80% of games where you aren't getting trilinear in all texture stages?

[nelg]

Leonidas]The problem with the current ATi/NV drivers: You cannot play 80% of games with a full trilinear/anisotropic filter. A 500-$-gfx cant do a full trilinear/anisotropic filter in 80% of games. Mmh, incorrect: The hardware can do this, but ATI/NV doesnt want you to use this potential.

As Dave mentioned before this is done for competitive reasons. Look at how your site somewhat misrepresented the abilities of ATI's cards (WRT their being able to full trilinear when selected in game). In this same vain how many sites would just set the slider on max quality and then bench without looking at the difference in filtering quality? If I am not mistaken two of the top three sites have already done this (AT and [H]). :cry:

[Xmas]

Iff IQ is not increased by doing trilinear on all texture stages, then why does it matter for those 79-80% of games where you aren't getting trilinear in all texture stages?
If... and who makes that decision? Some people are more sensitive to mipmap banding than others.

[3dc]Leonidas, Absolutely not true. With the ATI drivers and a registry setting, you can force max AF for all texture stages.
Which quite underlines Leonidas' point that ATI/NV don't want you to use this potential (for benchmarking purposes, I should add). Else it would be in the driver panel, not some undocumented registry setting.

As Dave mentioned before this is done for competitive reasons. Look at how your site somewhat misrepresented the abilities of ATI's cards (WRT their being able to full trilinear when selected in game). In this same vain how many sites would just set the slider on max quality and then bench without looking at the difference in filtering quality? If I am not mistaken two of the top three sites have already done this (AT and [H]). :cry:
At least 3DCenter made an update to the article saying that applications can enable full trilinear on ATI cards. Maybe if ATI added some switch to the driver panel enabling users to select full trilinear, some review sites would actually come to realize that there are differences to filtering quality that turn any comparison into green apples vs. green oranges.

[StealthHawk]

Iff IQ is not increased by doing trilinear on all texture stages, then why does it matter for those 79-80% of games where you aren't getting trilinear in all texture stages?
If... and who makes that decision? Some people are more sensitive to mipmap banding than others.

The people who are most sensitive to mipmap banding, of course 8)

[Hanners]

I wasn't. But, how many games from 1999 and before actually display issues with the filtering levels available through the control panels?

Actually, the problem seems to be more with modern games, most of which still don't have the ability to set anisotropic filtering within the application. Halo is the most recent one I noticed problems with on certain textures when AF is forced via the control panel - A quick trip into rTool fixed the problem.

[cthellis42]

Which quite underlines Leonidas' point that ATI/NV don't want you to use this potential (for benchmarking purposes, I should add). Else it would be in the driver panel, not some undocumented registry setting.

Hmm, is now the right time to mention that those kindly folks who run proper and extensive benchmarking and publish them are EXACTLY the folks who know about these settings? Plus, while it would be nice for the IHV's to compensate, I think it's MORE important for the game developers themselves to be the ones putting in the full-fledged controls THEY want in a game, so selecting "Application" in the video driver's control panel is all that's needed ever.

[Fred da Roza]

Maybe if ATI added some switch to the driver panel enabling users to select full trilinear, some review sites would actually come to realize that there are differences to filtering quality that turn any comparison into green apples vs. green oranges.

I doubt that. Look at AT latest article on the 9800XT vs 5950. Anand didn’t even mention that trilinear filtering doesn't work with the 52.xx drivers on the 5950. He missed AA issues on the 5950 as well. And that was supposedly an IQ article. ATI should not put themselves at a greater competitive disadvantage so they can have the even higher moral high ground.