Doesn't anybody at NVidia have any sense....

[pam123]

I used to think ATI's Quake fiasco was the nadir but NVidia just went them one better ( or is it one lower ? ).
They did this in the 2003 benchmark they claimed was invalid.
http://www.extremetech.com/article2/...1086025,00.asp

[glc]

This is why I could care less about benchmarks - how something performs in the real world is what counts.

[pam123]

Agreed.
I'm wondering about the Doom 3 benchmarks now.

[Blakhart]

Sounds of desparation. I hope ati wipes them up till they can make a fast card that has image quality too.

[simwiz2]

I don't think we should jump to conclusions here. For one thing, NVidia would have to be pretty damn stupid to cheat and trash its reputation like that. IMO, for a company with reasonably sane management to make such a desperate move, they would have to have an inferior product. And, as has been shown from benchmarks with actual games, the FX 5900 is not an inferior product. So why would NVidia cheat?

It could be a driver bug or glitch. I haven't read the whole article yet because it kept going down, but having read the replies to it, it seems like the website is speculating.

Simply put, NVidia has no motive to cheat in 3DMark and risk their reputation. Would it really matter if they lost 3DMark while their card beat the 9800 series in many real games? The new card, combined with an incredible driver, gave NVidia the lead. To cheat when you are already ahead is idiotic, there is nothing to gain and everything to lose. It's like cheating on a test when you already know the subject well: you risk getting caught and you would have done well anyway. I cannot imagine NVidia cheating now. Maybe a few months ago, but not now.

[pam123]

We can wait for NVidia's explanation, for now it doesn't look good.

[Spyda]

while people who know what they are on about care about how they perform in games, the guy whos just getting a new PC just sees the bigger numbers eh..

[sleepypost]

yea unfortunatly companies are willing to deceive their faithful customers simply to get more money. its just corporate ethics at work. Not ALL companies are bad though (did any body read the LCD article in PC World?).

[mystvearn]

I think Nvidia is trying very hard to deliver the fx5900 on a 4 pipiline..Not that supperior.

[simwiz2]

Is there really that much of a performance difference between 4x2 and 8x1 pipelines?

[pam123]

Yes, as the games get more demanding it's a contest as to which solution will do the job.
The loser goes into chapter 11.

[sleepypost]

i thought that with the pipelines it didn't really matter what config they were in. i thought that either L shape 4 x 2 or straight 8 x 1 it is utilizing the bandwith in a different way. i didn't know that that was some sort of marketing strategy...

[morriswindgate]

Benchmarks are worthless anymore, the manufactures have manipulated the cards they provide or there have been rumors of the manufactures and software people manipulating the programs to make one brand look good and the other look bad.

[pam123]

What's going on is that NVidia stumbled during one of the graphics industry's periodic games of "last man standing".
Whoever comes out on top after the next year or so will have a top position in what is becoming a "mature" industry.
Until September of last year it looked like NVidia was set to be the one, that isn't true anymore.
The tech guys are fighting this one, pipelines, shaders, and drivers ,all have to be changed. Thus ATI's Catalyst crew improving drivers beyond all expectation and NVidia cheating on 3dmark 2003, after declaring it "worthless", because their card did badly on it the first time around.
The real world test is still paramount but make no mistake, it's not "just marketing".

[simwiz2]

Their card did really badly on 3DMark the first time around, but it didn't do so well on anything else either.

The 5900 could *probably* have beaten the 9800 in the benchmark without the extra "optimization" because the 5900, plus the real optimizations in the new drivers, put the FX5900 in the lead overall. If the 5900 won most games, but the 9800 won 3DMark, couldn't one suspect ATi of cheating?

[pam123]

The folks at extremetech who nabbed NVidia went back and checked the ATI card as well.
Nvidia was the one cheating.

[morriswindgate]

NVida is turning into a company that manufactures "HYPE" instead of product. Does anyone in their right mind think that a company can release a new outstanding product every 4 to 6 months.
ATI took it's time with it's latest chipset and got it right, but NVidia produced 3 socalled improved latest and greatest in that time frame.
All NVidia does is repackage a slightly improved chip add a catchy name and hype the hell out of it so that people will pay 350 bucks for basically the same thing they have.

[simwiz2]

I'm not saying ATi was cheating. I'm saying that if the 5900 won decisively in many games and the 9800 won decisively in 3DMark03, it would be suspicious, especially when you consider who was still part of the 3DMark project when it was finished.

[pam123]

It won till Tomshardware benched it at high quality, then the 9800pro won. A reversal of the numbers.
http://www6.tomshardware.com/graphic...x_5900-12.html
"..it would be suspicious, especially when you consider who was still part of the 3dMark porject when it was finished. "
This is what's known as an unfounded accusation.

[simwiz2]

LOL that result is ONE game and ONE detail level, which seems to be having problems with NVidia's drivers. The 5900 regains the lead VERY DECISIVELY in 4X FSAA and then keeps it. In almost all the other benchmarks at THG, the 5900 wins.

"This is what's known as an unfounded accusation."

Note that I said "WOULD be" "suspicious", not "ATi won that bench they are cheaters". This means I am talking about a hypothetical situation, not accusing ATi of cheating. As if your "I'm wondering about those Doom III benchmarks now" isn't an unfounded accusation? Considering that ATi has doesn't have a clean record either, if I were to accuse ATi of cheating it would be no more unfounded than your statement.

[pam123]

read the whole review.

[kissfan]

All I can say is, that after applying the latest Drivers to my FX5200, I got a 50% increase in 3d03 and over 25% in 3d01.

That is by far the most improvement in a driver update I have ever seen. But hey if it works whats wrong with it.

Both companies want to have the fastest card , and how they acheive it is surely their business, you are not being cheated, the scores will reflect how the card can perform if allowed to employ it's own optimisation methods.

just my 2p's worth

[Spyda]

all of Nvidias catchy names and hype have done nothing for me, i still wouldnt touch an FX card

my next card will probably be a 9800, because by the time im ready for a new card it will have a 'real' price and not some number times 3

[mystvearn]

Spyda is right. I think Nvidia has no ideas staying on a piplein technology while ATi has 8. This is the last card I am going to buy from Nvdia unless they can prove thmselves winners by a clear margin.

[Spyda]

i think the Ti4*00s were the last of the good efforts, they've lost the plot a bit now

[mystvearn]

Well-no one is immortal-can stay on top forever. Which is good news to consumers when there is rivalry