NV40 3DMark2003 screenshot (possible)

[Mrstickinit]

I found this over at Futuremark's site. This could very well be Photochopped, or not? Here is the original link to the site from which the screenshot came. http://news.mydrivers.com/pages/2004...1531_17315.htm

[Mrstickinit]

Here's the screenie.....

[pkrew]

Hard to tell, but I have a score just like that one on a 9800pro. It would be nice to see the individual scores and the system specs

[Mrstickinit]

Originally posted by pkrew
Hard to tell, but I have a score just like that one on a 9800pro. It would be nice to see the individual scores and the system specs

You have a 12,000 score in 03 on a 9800pro?? Maybe on one of the individual tests, but surely not the composite score.....

[pkrew]

LOL, you can't get 12k on indiviual test. It was a glitch score, but that's my point there are several of those out there.

[Mrstickinit]

LMAO, I didn't mean 12,000 on an individual test. HeHe. I meant that you possible got a similar score on one of the tests like in the screenie.....

[eva2000]

Originally posted by Mrstickinit
I found this over at Futuremark's site. This could very well be Photochopped, or not? Here is the original link to the site from which the screenshot came. http://news.mydrivers.com/pages/2004...1531_17315.htm

what's new > 12,962 http://www.xtremesystems.org/forums/...388#post417388

[LilGator]

That was a bugged nature score... the rest on here are all consistent...

[Arkangyl]

I'm 80% sure that's photochopped, look at the 2 and the 5 in the 12,510 score in the big box on the right, the 2 looks a bit misplaced and so does the 5, they look a little too close to each other relative to the other numbers.

[Orak]

Originally posted by Arkangyl
I'm 80% sure that's photochopped, look at the 2 and the 5 in the 12,510 score in the big box on the right, the 2 looks a bit misplaced and so does the 5, they look a little too close to each other relative to the other numbers.

Yes, it can be photoshoped... but, ill wait for some trusty scores are out

[macci]

I dont think that its a real score at all.
The GT1 at 450+ FPS just requires too much from the rest of the system - something like P4EE @6GHz might do it
Also how come the system info part of the 3dmark screen has been blocked there?

Also if you do the maths..
10008 x 1.25 = 12510
I think that someone just wanted to be a clever boy and faked a 'NV40 stock 3DMark03 run' - it "happend" to be excatly 25% above the current top hardware

[r3b0rN]

Originally posted by macci
Also if you do the maths..
10008 x 1.25 = 12510
I think that someone just wanted to be a clever boy and faked a 'NV40 stock 3DMark03 run' - it "happend" to be excatly 25% above the current top hardware

Omg Macci nice find man...

[Geforce4ti4200]

how cpu intensive is game one anyway? wont a faster gpu still increase the fps in that first dx7 test?

[macci]

Of course the GPU speed will give an improvement there but the score just doesn't make any sense. It seems that the GT1 is getting clearly the biggest boost here - and its a DX7 test (most system speed dependent)!

Here is what an average system (lets say 3.2GHz P4) w/ very highly clocked FX5950 would get:
GT1 = 260
GT2 = 64
GT3 = 52
GT4 = 56

Now here is what the 'NV40' gets - I'm assuming that its useing a normal CPU power as its supposed to be a stock clock score (so ~3.2G P4 system speed)

GT1 = 453FPS 74% boost over NV35!?
GT2 = 83FPS 30% boost
GT3 = 72FPS 38% boost
GT4 = 71FPS 27% boost

74% boost in the dx7/system speed test and only 27-38% extra speed in DX9 GPU dependent tests?
Doesn't make much sense does it?

[zakelwe]

Originally posted by macci
Of course the GPU speed will give an improvement there but the score just doesn't make any sense. It seems that the GT1 is getting clearly the biggest boost here - and its a DX7 test (most system speed dependent)!

Here is what an average system (lets say 3.2GHz P4) w/ very highly clocked FX5950 would get:
GT1 = 260
GT2 = 64
GT3 = 52
GT4 = 56

Now here is what the 'NV40' gets - I'm assuming that its useing a normal CPU power as its supposed to be a stock clock score (so ~3.2G P4 system speed)

GT1 = 453FPS 74% boost over NV35!?
GT2 = 83FPS 30% boost
GT3 = 72FPS 38% boost
GT4 = 71FPS 27% boost

74% boost in the dx7/system speed test and only 27-38% extra speed in DX9 GPU dependent tests?
Doesn't make much sense does it?

Although I wouldn't normally argue with macci's analysis I have had a few beers this afternoon so would like to throw up

1) Although the GT1 score does look AWOL as it is the only test that is cpu partially dependent, nvidia complained that GT1 was single textured ( mainly ) when it came out which did not suite their 4x2 cards. Now if 16x1 rumours are true then maybe it will be a lot better ?

2) What are the ratio's like ? Does the ratio of GT2 to GT3 to GT4 look ok ? If it does how did they accurately guess these ratio's but then go nuts on GT1, that would be a big mistake? If GT2 to GT3 to GT4 is rubbish then I subtract my first point.

Regards

Andy

[macci]

The funny thing here is that EVERY GT result is pretty much exactly 25% higher than on my 10008 score (give or take 1%)

So basically it could be true IF:
A) hes running a 4.5GHz P4EE
B) the NV40 at stock speed is indeed exactly 25% faster than the R9800XT @657/471 in every Game Test
c) NVIDIA has really improved there DX7 speed

[zakelwe]

Originally posted by macci
The funny thing here is that EVERY GT result is pretty much exactly 25% higher than on my 10008 score (give or take 1%)

So basically it could be true IF:
A) hes running a 4.5GHz P4EE
B) the NV40 at stock speed is indeed exactly 25% faster than the R9800XT @657/471 in every Game Test
c) NVIDIA has really improved there DX7 speed

Yeah, maybe I am reading too much into this and they just decided to bench windows calculator to get the score !!!!! Given past history this is almost certainly the case. However ....

Playing it as true still though, if C is true then they did so at the cost for DX9 as his GT4 score is lower than his GT3 ( DX8 ) where as yours is higher ( as is FUGGERS and OPPS ) .

This would paint a picture of nv40 being vastly better in single texture DX7 games, OK in DX8 and needing some work on the drivers in DX9.

Whatever, I'm looking forward to 2001 and 03 scores come Tuesday to see what's what, if any of the major web sites are still using it

Regards

Andy

[OPPAINTER]

Is this the real Screen shot that every site is talking about showing that the NV40 can do 12k in 03,,, or is this some screeny made up by someone after the rummor that the NV40 could do 12k?

OPP

[Jrocket]

They probably just changed the settings to a lower resolution or something to make it a higher score.

[Soulburner]

Originally posted by Jrocket
They probably just changed the settings to a lower resolution or something to make it a higher score.

Could be, but the pic says 1024x768, but that could be edited too I guess.

[Mrstickinit]

Originally posted by OPPAINTER
Is this the real Screen shot that every site is talking about showing that the NV40 can do 12k in 03,,, or is this some screeny made up by someone after the rummor that the NV40 could do 12k?

OPP

Who knows OPP.... As I said in my original post, I pulled this off of Futuremark's website last night and it could entirely be fake. macci makes an interesting point in the fact that the score is 25% faster than his, however, it very well be a correct score.

Has anyone gone to the Korean? website in the link I provided and translated? I see that there are comments on there about individual scores but cannot make out what they say.

nVidia may have pulled out all of the stops on this card folks. Maybe they are tired of ATI kicking their a$$es in the graphics war and have played a trump card. I guess we'll find out in a few hours.....

Let me ask this though. Over the last few days and weeks, we have seen all kinds of claims on what the NV40 may or may not be capable of, including clock speeds, pipelines, Pixel Shader counts, card layout, molex power connectors, etc. Where are all of the r380's specs? How come we're not hearing about ATI's latest and greatest? Yes, we have heard a few things about it, but only after a claim is brought forth about the NV40 in a refute attempt. Is ATI silently shaking in their boots about the NV40, or are they laughing behind the scenes knowing that their next offering will once again be the best and the brightest......?

[OPPAINTER]

I'm just thinking that if this is the screeny that the 12k rummor was started by then maybe the card isn't all that great,, if this is a fake screeny.

OPP

[Mrstickinit]

Originally posted by OPPAINTER
I'm just thinking that if this is the screeny that the 12k rummor was started by then maybe the card isn't all that great,, if this is a fake screeny.

OPP

I dont believe that this is the screenie that started it OPP. The rumor was started after a story was posted at TheInquirer website that said the NV40 was pushing 12,535 3DMarks. Here is the link to that story http://www.theinquirer.net/?article=15169

[zakelwe]

Lots of unkowns here, the 12 535 is the score nvidia were pushing around but nobody knows which driver set they were done with. If you use Futuremarks "good" 52.16 you lose 700 points going from 53.03 and that is with a 5700U where you are only scoring 5000 and not 12000, so depending on the driver set a real certified score might be 12.5k minus 1k , assuming 12k is accurate.

On the other hand the 60 series drivers might make it all back without optimisations ??? Who knows ?

And that is not taking into account Rivatuner which will make a big + on GT2, GT3 and GT4 .. so there are too many unknowns like I say, best to see what the first testers do and what driver set they are using.

What I am interested in is 3dmark2001 with this card, if it can be used with the "optimised" 44.03 drivers and it really is very very good in DX8 then nature is going to be massive.

The one thing you can say about nvidia is that there are plenty of driver permutations, that is why I love benching them so much.



Regards

Andy

[macci]

Its possible that the NV40 will pull a 12k stock but the score SHOULD NOT look like that one posted on this thread. GT1 should be lower and GT2-4 higher as those are more video speed dependent.

System speed (CPU/RAM) is the limiting factor in several parts of the GT1 so video speed boost will not make a huge improvement on that one.
Theoretical 20% boost in video card speed would give something like 20% boost in GT2-4 and 5% in GT1. in the 12.5k pic its just the opposite - and it makes no sense