NV at it again....

[speeduk]

NV are at it again (or rather caught at it again). New 3dmark 03 patch reveals all...

5950ultra build 330: GT2 43.9fps GT3 34.5fps MN 34.1fps
5950ultra build 340: GT2 36.9fps GT3 30.3fps MN 25.1fps
Overall 5800 - 5000 with latest version...

<A HREF="http://www.nordichardware.com/article/2003/3DMark03_340/" target="_new"> Click</A>





My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[speeduk]

Oh and the ATI benchmarks are almost the same with the previous version. Just some 1-3% variations.

My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[rain_king_uk]

Cool thanks for the link - I came here to see if anyone had tested the new build yet.

 

[rain_king_uk]

Looks like the ATi variances are within the margin of error for the test as some of them are plus and some minus but only small amounts.

Let's see what THG has to say about this - my money is on them not kicking up a fuss since it shows nVidia in a bad light. *shrug*

This makes you wonder what cheats nVidia have used to increase their scores in actual game benchmarks too.
<P ID="edit"><FONT SIZE=-1><EM>Edited by RAIN_KING_UK on 11/12/03 06:36 AM.</EM></FONT></P>

 

[speeduk]

Does this change anything though? Nvidia don't listen to anyone in regards to their cheating. It seems like their next driver release will only add back the cheats so they gain 800 3dmarks......

No point in gettin the update if you have an ATI card 'coz you get the same score, but if you have an Fx card, then you probably wont get it anyway because it makes your score lower lol.......

My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[rain_king_uk]

I'm looking forward to someone actually suing NVidia and testing in a court of law whether their cheating is misleading customers in an illegal manner. I think it probably is - we just need a lawyer willing to put together a class action suit and some willing NVidia customers.

<P ID="edit"><FONT SIZE=-1><EM>Edited by RAIN_KING_UK on 11/12/03 06:44 AM.</EM></FONT></P>

 

[speeduk]

Its because of 3dmark and the DX8 game benchmarks that we look at the FX line in the same light as the 9800. The 5900 = 9800, the 5900u = 9800pro and the 5950U = 9800XT.
If these optimizations were found out earlier and removed, then people would think the FX suck, like they do in the latest games like max payne 2 compared to a 9800pro.

So basically a 5950u gets you 5,000 3dmarks and 60fps in max payne 2 and most new DX9 titles while a 9800XT gets 6,000+ and 70-90fps in the latest games...



My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[speeduk]

It just proves how good the 9700pro really was, and still is. It matches the top cards from NV easily, especially on a non optimized benchmark like the new 03.....

My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[Ion]

All FX owner shouldn't worry about 3dmark 2003, the next NV drive release will cure it......

 

[Ion]

While we at 3dmark, check out this 3dmark score!

<A HREF="http://service.futuremark.com/compare?2k3=1532116" target="_new"> 9131!! </A>

 

[rain_king_uk]

<A HREF="http://www.beyond3d.com/articles/3dmark03/340/" target="_new">http://www.beyond3d.com/articles/3dmark03/340/</A>

 

[rain_king_uk]

At one we asked Derek how this sat with the optimisations guidelines that were given to press by NVIDIA, specifically the guideline that suggests "An optimization must accelerate more than just a benchmark" To which Derek's reply was "But 3DMark03 is only a benchmark".

Is he that much of a dumbass not to see by that Futuremark mean application specific optimisations are not allowed? What a [-peep-] retard.

 

[Ion]

He is just retelling whats been told to him by his boss(unless he want to get fired). :tongue:

PS. that matrox card is sexy in B3D review.

 

[speeduk]

I think its safe to say that the optimizations found in 03 exist in DX9 games too. Is that a good thing or a bad thing? Time will tell I guess...

My system spec: Fast PC<A HREF="http://service.futuremark.com/compare?2k1=7000747" target="_new"> 3D-2001 </A>
"It's not the spoon that bends, it's only yourself."

 

[Ganache]

Let's hope tomsH makes an article that talk about this issue. This will help them show their are not Nvidia monkeys. And maybe screw NV again. I really starting to hate Nvidia.

 

[TheGreatGrapeApe]

That's one SchweEet comparison! Nice IQ captures.

I was originally thinking, well perhaps this is just 3Dmk03 having problems withy the run-time compiler, but still giving similar IQ. But obviously that ain't the case.

Especially since it's mainy the DX8/8.1 tests that are at issue.

Can't say I'm surprised, but I was hoping that we were past this $hit!


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK

 

[Ganache]

bump so this does not go unnoticed.... I think it's a serious issue worth looking into.

 

[Vapor]

What about suing Apple?? They almost do the same thing by only providing unfair and heavily G5-biased benchies. nVidia should not be sued IMHO, in theory they could claim the optimizations not as a means to gain a higher score but rather to provide a faster playback of the benchmark--for enhanced viewing. I had another idea of why they shouldn't be sued, but I forgot it :frown: .

Just wondering, does this 340 build affect the Omega 'optimized' drivers?? Just a thought, but Futuremark may be partnering with ATi and jointly making nVidia's scores drop drastically. The 1-3% ATi drop would make sense too, as they would try to hide any association with ATi by also lowering the score, just not by a lot.

Damn Rambus.

 

[TheGreatGrapeApe]

Vapor your conspiracies are brutally simplistic. First, nV is just as much a partner now as ATI is.

Second, despite Apple's laughable claims, they are based on independant tests that Apple didn't floptimize for the way nV obviously did here.

And lastly, it doesn't matter how nV sells it. The end result isn't the same, they've gone against their OWN audit standards. Anyone wanting to sue nV would be able to use nV's own literature and rules against them. The most major failing of their rules, do not compromise image quality.

I still hold out a bit of hope that this may simply be an issue with the run-time compiler, but that's a faint hope.

And your theories are even more laughable than OJ's.

BTW, the Omegas don't offer much in the way of benchmark immprovements especially not in 3Dmk03, so you're barking up the wrong tree there too.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK

 

[coolsquirtle]

I dont see why anyone is pissed off right now. Sure nVidia optimized the cards AGAIN, but they did it clearly without losing IQ or produce a lesser Image than the ATi cards. What's wrong with that? Are you saying we should keep away things that can make our Videocards run better? I bet if ATi released a patch that increases the Score in any benchmark all of a sudden and u can't find any difference in IQ, not one of u is going to avoid it.

This is like the "boy cried wolf" story. nVidia cheated once, now you guys are doubting EVERYTHING that they do.
It like two S2000, one blue, one pink in a race and both running with out mods, now the blueS2000 was more fine tuned to fit the track before the race, so that blue S2000 won, now they moved on to the other track and that blue S2000 didn't fit with this new track so it lost, now the pink S2000 that didn't do anything yells at the other driver saying that it was cheating.

the Pink S2000 clearly could have been fine tuned but the driver chose not to.

Seen the resemblence? It might be a crappy comperason but that's what's happening right now, ATi very well can also optimize their cards for 3Dmarks but they know they have the upper performance advantage so they didn't. nVidia on the other hand had to inorder to be on par with ATi.

THat's my 2.6 cents worth

RIP Block Heater....HELLO P4~~~~~
120% nVidia Fanboy
bring it fanATics~~ nVidia PWNS all!
SCREW aBOX! LONG LIVE nBOX!!

 

[rain_king_uk]

Firstly, Futuremark clearly want their benchmark's to represent the performance of cards without application specific optimizations. The reasoning for this is sound - NVidia only optimize for the most popular games therefore if you allow application specific optimizations in 3dmark then that isn't a fair representation of the raw performance you are going to get in games in general (including the ones NVidia haven't optimized for).

Secondly, I would argue (and I think the evidence backs this up) that the optimisations that NVidia have implemented in many cases have reduced image quality and/or are not showing precisely what the application's developer intended - this is not acceptable to me.

It seems that nVidia's "unified compiler" is just a whole bunch of application specific optimisations that replace the application code with nVidia code which runs faster but in many cases reduces IQ or does not show what the devloper intended. (I wonder if the bigwig managers at NVidia even know/understand this or whether they are being sold a lie by the driver developers as much as we are).

Personally I don't think any driver "optimisation" that is application specific is a good idea at all and futuremark are completely correct in disallowing them in their benchmark.

I hear there are already some new NVidia driver's floating around that turn the "optimisations" back on. *SIGH* When will they learn? I guess only when people wise-up and drop them like the POS they are.

<P ID="edit"><FONT SIZE=-1><EM>Edited by RAIN_KING_UK on 11/12/03 11:29 PM.</EM></FONT></P>

 

[coolsquirtle]

again that's just ur assumption, no one had discovered ANY IQ decrease with the new forcewares.

RIP Block Heater....HELLO P4~~~~~
120% nVidia Fanboy
bring it fanATics~~ nVidia PWNS all!
SCREW aBOX! LONG LIVE nBOX!!

 

[rain_king_uk]

There's IQ decreases documented in an article linked to in this very thread.

 

[phial]

again that's just ur assumption

just like every single thing you said





ATI didnt cheat because they already had teh performance advantage? how do you know that? perhaps they didnt cheat because they didnt wanna taint their brand name like nVidia has so gracefully done this last year or so

-------


<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>

 

[TheGreatGrapeApe]

Yeah CoolS read the <A HREF="http://www.beyond3d.com/articles/3dmark03/340/" target="_new">Beyond3D article</A> (linked to above by RAIN), very revealing, and ALOT of IQ differences, surprisingly mainly in the DX8 areas not DX9 (WTF!?!) which is where they already outperform many of the Radeons. Now does that DX8 performance need questioning as well as their DX9?

I'm willing to let them go on the Forceware as long as it does provide equal IQ, but really it isn't in this case. The only problem I had with the Forceware drivers was having to wait for the nV engineers to optimize the run-time compiler to acheive similar results that were appearing without ATI needing these same tweaks. That's one reason why 3Dmk needs to be optimization free for the most important benchmarks IMO. You could always run two runs, one with and one without optimizations. But you can't just allow these kinds of issues to go unchallenged.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK