As Carmack has pointed out, and everyone seems to ignore, you CANNOT compare ATi to NVidia ps perf directly, it's impossible. ATi does everything at 24bits, reguardless of what you tell it. NVidia can do it at 12 int, 16 float, or 32 bit float, but NOT 24bit float to match ATi. So, the NV cards are either running higher precision and therefor doing more work per pixel, or running lower precision and putting out lower quality images. There is no direct comparison possible therefore it's really silly to get bent out of shape over 3DMark 2003. Wait for games to come out, then see how they do, and compare images.

U guys should check this article http://www.hardocp.com/article.html?art=NDgy

Here's a good one.....

"FutureMark Statement:
Futuremark now has a deeper understanding of the situation and NVIDIA's optimization strategy. In the light of this, Futuremark now states that NVIDIA's driver design is an application specific optimization and not a cheat.

NVIDIA Statement:

NVIDIA works closely with developers to optimize games for GeForceFX. These optimizations (including shader optimizations) are the result of the co-development process. This is the approach NVIDIA would have preferred also for 3DMark03.

Joint NVIDIA-Futuremark Statement:

Both NVIDIA and Futuremark want to define clear rules with the industry about how benchmarks should be developed and how they should be used. We believe that common rules will prevent these types of unfortunate situations moving forward.

Translation VIA HardOCP: FutureMark reneges on previous statements and confirms NVIDIA was not cheating on their benchmark and NVIDIA will not take a legal action against FutureMark that would bankrupt them. All about the $, but that is just our opinion. Still this does not change our thoughts on the FutureMark and their benchmark. Editorial coming..."

I have NO respect whatsoever for Nvidia. None. What a bunch of babies. Too bad that the whole hardware and gaming community knows it's cheating. Even Carmack in his definition says so as well as Tim Sweeny. Boy, all this does is make Nvidia look even worse. Bunch of whiners.

Dell Speaks...
02-Jun-2003, 16:45.59 Reporter : DaveBaumann
...and usually their suppliers listen.
It doesn't seem that often that Dell speaks out on such issues, but being one of the largest OEM's its usually in the best interests of their suppliers to listen. Today Futuremark have put up a quote of endorsement from Dell on its much battered 3DMark03 benchmark.
"Dell uses many tools to evaluate system and graphics subsystem performance. We believe 3DMark03 is a solid synthetic graphics benchmark that covers a wide range of usage models and complements application-specific testing. Synthetic benchmarks like 3DMark03 help to differentiate graphics subsystem performance characteristics of both high end and lower end cards by utilizing sets of tests with varying degrees of graphics complexity. Entry level cards will be able to run at least one simple test to be used in comparisons for those interested in basic 3D functionality. Additionally, those interested in leading edge technology with be able to make graphics hardware comparisons with a range of tests using the new APIs, shaders, rendering techniques, etc. Dell believes 3DMark03 is a versatile tool that allows a fair comparison of today's wide range of 3D graphics solutions."

And this is why Nvidia cares about this whole issue and trys to b.s otherwise. OEMS still use this benchmark. And Nvidia knows it, like it or not. And if Dell uses it. You know the others do too. The great fear of Nvidia right now.

http://www.darkcrow.co.kr/image/news/0305/CheatingDrivers.jpg
http://www.visiontek.com/

http://www.rage3d.com/board/showthread.php?s=6c2549e3f04bb820896c17...

Chk these out. It's funny as hell. Funy shit about Nvidia cheating.

And the funny thing about what Carmack said about all this OR lack thereof is that he doesn't realise that the required minimum specification for DX9 full precision is FP24! Not FP12,16. LOL. Unless he knows it and did not want to mention it. If he did, that would be saying Nvidia did cheat. And the man has to work with Nvidia in the future and has always stood by them. Best to keep ALL SOME FACTS QUIET. lol. As they ran those 3dmarks tests in question at FP12, FP16 not the required FP24 or FP32.(SHADER TESTS. Nvidia's weak spot)
That is not optimization ladies. That is cheating. Every GPU company knows the required specifications of dx9. To say otherwise is lying. Because the fx line of chips shader engines run slower then ATI's. And running them in less precicion will increase performance, but degrading quality. YOU HAVE TO RUN IT IN THE REQUIRED SPECS TO BE VAILD. And look at the very few game and syn tests the 9800pro won over the 5900 Ultra. They were games and tests that...you guessed it, uses lots of vertex and pixel shaders. The 5800, 5900 weak spot. This is why the shader tests for 3dmarks were targeted by Nvidia to cheat in. That is one of their cheats. Beyond 3d and TechReport show this and prove it. Naughty, naughty, naughty Nvidia. You have been caught big time. And by Carmacks definition of optimization, it shows ATI did not cheat. They ran the specifications and the outcome was THE SAME. Not for Nvidia. Artifacts galore when the cahnage the rendered camera path. Nvidia's 2nd cheat.
Funny also how Carmack does not bring that dubious other "tweak";) in his response. No mention of clipping planes. Loss of clearing buffers...Hmmmm.;)
This is a little part of what was said at TheInquirer....

"There is no way that reducing the workload on their GPUs, replacing shader code and using custom clip planes to artificially increase their scores is a legitimate tactic, and it should be ashamed of its actions. Even such industry heavyweights as Tim Sweeney, one of the lead developers on the Unreal series; and John Carmack, Doom III mastermind, have weighed in on the subject, again reiterating that an optimisation is not considered valid unless the end result is exactly as it was intended after optimisation. In Nvidia’s case the end result was similar, but not the same, but in ATi’s case the end result was exactly as it was meant to be."
Carmack in his comment about this whole issue left ALL the juicy cheat details out also. lol. Don't piss off one of the hands that feed you i assume. But he should have addressed the deliberate cheat. He just did off the mark. Not pointing fingers at Nvidia. Even though his very definition of cheating defines what Nvidia did. And his definition of optimization defines what ATI did. Read up on it. And Tim Sweeny's take too. Another heavy weight in this area. The lead programmer and designer for Unreal.......

Nvidia should be lauded for mocking this pointless 'benchmark'.

It proves little as the engine is not used in ANY application in real life.

All this hot air and bandwdth over a SYNTHETIC piece of five minute eye candy is BORING me to TEARS.

/me turns off NTK mode

you are right you cannot compare ATi to nVidia to do so would be insulting to ATi....

YOU MEAN TO TELL ME THE ONLY WAY YOU (NVIDIA) CAN COMPETE WITH YOUR VIDEO CARD IS TO HACK THE IQ AND PERCISION OF YOUR SHADERS AND VORTEX PROCESSING UNITS.

Funny that nv3X is really a great DX 8.1 card that only introduces you to DX 9 featureset while ATi actually has a card (R3X0 series) that is a great DX 8.1 AND 9.X performer.

Well www.hardware.fr has a review of the new 5600 Ultra today, using Det 44.03 drivers.

They test the High Dynamic Range test and Humus Mandelbrot rendering program. Both were still very slow. No data for an older driver set though...

HDR (1024*768 32 bits, AA 4x) -
5600 Ultra - 1.78 fps
ATI 9600 Pro - 10.5 fps
ATI 9500 Pro - 13.9 fps

Mandelbrot (1024*768 32 bits) -
5600 Ultra - 9 fps
ATI 9500 Pro - 45 fps
ATI 9600 Pro -32 fps

Link to a translated version - http://216.239.37.104/translate_c?hl=en&langpair=fr%7Cen&u=http://w...hardware.fr/articles/468/page4.html&prev=/langua...[]

Like I said, shader 2.0 on the NV30, NV35 and NV34 is awful. There is no excuse. The only reason the improvements come is because the shader path is comepletely replaced and inaccurate.

Does anyone where to get the Shader Mark benchmark? Can't seem to find it anywhere.

Like I said, check Shadermark or 3dmark, and it will tell you exactly what Carmack said. NV30-NV35 shader 2.0 performance is horrible. Check here.

http://www.hardocp.com/image.html?image=MTA0NjgwOTI1NEdRU0hrZlkyeDR...

It's so ugly they didn't even make a bar graph. There is up to 400% better performance relative to the 5900 from the 9800 Pro. I'd much rather have a card that is future proof than one that can get a few fps more in some current games.

Then NVidia did some "cheating" for this benchmark too. ANd still lost to ATi.

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

I'd like to see different drivers run on this benchmark.

And the debate goes on.

oh BTW, remember back then when Nvidia was still making TNT and GF1 cards? each driver they release increase their perfomance by like 10-20% atleast, now it makes me wonder if they achieved that by cheating in every game also.

back then the 3d graphics look like shit, 16bit color, low poly, low res texture, no AA, we prolly won't even notice a damn thing even if they cut corners in their drivers.

Nvidia really needs to come out, say they are sorry, promise to never do it again, and do some token fires to anyone and everyone who made the choices to do this. While I am sure managment would have been in on it too, getting ride of some of the lower guys could really do a possitive PR trick for them. Otherwise the next round of card wars might not be desided on graphical power, but bad PR.

you all must have SUPER DUPER VISION cause I can't tell crap from the screen shots.. they all look the same to me.. Looking at the Nvidia ones, and then the ATI ones.. I can't tell the difference..

Not only has Nvidia been panicking and cheating since ATI released their 9700 Pro, now Nvidia is trying to "force" people into buying cards based on their technology as witnessed by the deal they struck with EA not that long ago. I guess the days of legitimately trying to win over customers through genuinely superior hardware (and drivers) are over. A sad day for gaming on the PC indeed.

That fact remains that in shader 2.0 (non vendor specific) the NV35 loses big time, synthetic benchmark or not.

It's funny how everyone claims "who cares," and "the benchmark is synthetic" when John Carmack had already said NV30 is half as fast as R300 in DX9 ARB2. So what makes a synthetic benchmark not worthy?

I remember reading a post from "The Carmack" about a year ago, maybe less, in which he said that Nvidia was basically behind the tech curve for their next cards because of the amount of time they had to invest in the chipset for the Xbox. I didn't really think much about it till recently, but I am thinking he could have been right.

Now this is a stretch of logic reaching levels near "If you use drugs you are supporting terrorism," but if all this ends up taking down Nvidia in the end, you could almost point some of the blame to Microsoft. If I am correct they where having problems collecting their share of profits from the sales of the Xbox too.

Yes, you may flame away. :)

Agree with some of the posters below. Who cares about driver optimizations in 3dmark. The direction that Benchmark took in it's latest delivery was a sad state. The bench is ahead of it's time as far as measuring 9+ months of future games. proof: GF 3's tank and idle at 1-2 fps during that test, yet they play "fine" every other current game. The difference in scores with similar cards from ATI/Nvidia shows us that something is WAY off. Ideally the bench should have gathered up like the last test games to be released 3-12 months later and gotten snippets of up and coming real games like HL2 or DIII or any graphic intensive game to be released. Everyone wins. The gamers get to see an example of future tech. Industry wins because hype and free commercials for new games, benchmark companies win because of actual real use in applying logic. Graphic card makers win because you have easy goals to shoot for. OEM's win because they can say with confidence how a card will produce on their systems. meh. For once I agree with Kyle on this, the benchmark should not be used period in evaluating a card's merits.

3DMark should be abandoned as a method for testing video cards.

A better benchmark would be an aggregate score of performance on various 3D games and other applications.

Futuremark is on the brink of extinction.

I will base the purchase of my next video card on the quality of drivers, performance on 3D games, video quality and bundled software. I will totally ignore synthetic benchmarks.

Heh-Heh...;) Judging how pleased some people are upon learning how nVidia's been attempting to cheat their perceptions of its products I can understand why con artists do so well these days.... I guess it's really true that there's a sucker born every minute...;)

I can't agree with the opinion that it doesn't matter blatant cheating is going because 3dmark2k3 is just a benchmark. I can't say for some of you, but when a new spiffy video card comes out, I am very curious about the speed of its new feature sets, not just the performance on current apps. It just happens that the big feature on DX9 level hardware is ps/vs 2.0. And which current app uses those features? AFAIK, only 3dmark2k3 and (perhaps?) shadermark. Sythetic Benchmarks. Just how are we suppose to evaluate those new features without those tools? How do we tell the qualitative and quantitative differences in rendering a part from the last generation of cards? Wait another 2 years for a game using it to come out? Nvidia know 3dmark2k3 shows the new features on their cards as underperforming, and that's why they cheated.

Just say no...
To 3dmark 2K03.
It serves no real world purpose

If ATI's optimizations are indeed not cheats, they should keep them.

Screw futuremark and 3dmark 2k03.

I think the whole thing is funny.

Nvidia and ATI both made optimizations. Did they cheat? Sure... but nvidia's cheat was a little more "cheater-ish" since they took their optimization *way* too far. Pretty smart of them... but dumb all at the same time.

I say: WHO CARES

Companies have manipulated benchmarks since they first appeared many years ago.
It's just getting more complicated. Both ATI and nV are guilty.

I'll buy whatever card offers the most value for the dollar and runs the games I play the best. Nobody plays 3DMark. People shouldn't be getting their collective panties in a bunch over it all. Pretty soon DX9 games will be out and we can benchmark those. I'm sure there will be game-engine-specific optimiztions at this point as well. Let's just try and live with this behavior, mmmkay?

In the meantime, like I said, WHO CARES if optimiztions are being done in drivers to make synthetic benchmarks perform better. If I was a driver software engineer, and my boss told me to make it faster, I would pull out every trick in the book to make it so.

PPL should go outside and think of something else to talk about....all this ATI/nV sucks because.......... crap is boring me!

To me, this increases ATI's respectability due to the fact that they owned up to a little optimising, rather than NVidia claiming it was an accidental driver fault.
I don't see a problem with driver optimising in general, but when they're application specific to improve their public image, that's a different story.

ok attention all nVidiots!!!! If futuremark's benchamark 3dmark 2K3 is worthless( like you say it is) why did your precious nVidia WASTE RESOURCES that could have been used to improve their drivers on it. So what if your card is faster at 4xFSAA (supersampled) it looks WORSE than ATi's 2x rotated grid multisample FSAA.

Oh great nVidiots of the world please explain to me why does nVidia promise claim the "Dawn of Cinematic Computing" while all this time ATi has delivered what nVidia keeps promising..

Aside from nVidia's blatant differences in the rendering of the turtle between the two 3DMark builds, anyone else notice how nVidia and ATi render the surface of the water very differently? This isn't shown in the B3D diffs, which only compare old and new 3DMark builds. Best way to see it is to load the nVidia and ATi "build 330" images into two different tabs in your browser (or two different windows one on top of the other if you're not using Mozilla), then switch quickly back and forth using keyboard shortcuts.

The images are all supposed to be of the same frame. And yet, the two cards render the water in a slightly different position. Just curious whether anyone has any ideas as to why this would be happening.

The suckiest thing about this whole mess is that nVidia has made 3DMark03 seem like an invalid "useless" benchmark to a lot of people, which is probably exactly what they were shooting for.

If I were Futuremark, I would seriously consider legal action.

Didn't 3.30 show some pretty significant framerate improvements in other benchmarks too, like UT2003 and Q3? Any chance of them pulling of cheats like this with those too..?