NV40 3DMark 2003 scores revealed -theinquirer

[martrox]

I can confirm that Far Cry @ 1280x1024, 4x AA runs slightly smoother on the NV40 than on the Radeon 9800XT with no AA. But hey, with 1200Mhz effective memory, what did you expect?

Uttar

Got Far Cry yesterday, am playing at 1280X1024, 4X FSAA and 8X AF, everything maxed, and it's very smooth on my 9800Pro 128 @ 430/360 on a P4 2.6 @3.4........

 

[digitalwanderer]

Keep in mind that due to a doubled register pool, the maximal advantages of the partial precision hacks in 3DMark03 are divided by (at least) 2.
Theorically, this means that if a NV3x had a score of 4000 with 52.16 and 5000 with newer drivers, a NV4x would in pure and simple THEORY need a score of 4500 with non-cheating drivers to get 5000 with cheatin drivers.

Keep in mind this is pure and simple THEORY. It is in no way truly representative of real differences between cheating & non-cheating drivers for the NV4x, because:
1) Legit optimizations are possible (although not many of them IMO, unless NV40's current compiler is godawful, or NVIDIA decides to use several hundreds manhours on optimizing 3DMark03 for the NV4x).
2) Other things than partial precision can be used to improve performance, although NVIDIA has already promised not to use them anymore, both to the public and privately to FM AFAIK.
3) The NV3+ chips (NV35, NV36, NV38) are not capable of FP32 MADs at full speed.
4) It's a different architecture (plus, saying the register file is "doubled" is not sufficient because several other, highly related things are also changed)

So if that is sort of accurate and I am doing my math sort of right then the 12,535 should be more like a 11,281...and Fuad mentioned that the R420 is hoping to crack 10k.

Sounds like a shooting match, I'm excited! 8)

can confirm that Far Cry @ 1280x1024, 4x AA runs slightly smoother on the NV40 than on the Radeon 9800XT with no AA. But hey, with 1200Mhz effective memory, what did you expect?

How does the AA look on the nV40 compared to the 9800's AA? (I'm not asking for technical details you'll note, just your impressions/opinion. )

At least we will see some benchmark, for sure, on the 13th

Nope.

You are an evil, evil man! What are you talking about? Isn't the launch date the 13th, or do you think it might just be one of them paper launches with no product available.....








....UNTIL JULY! 8)

 

[Tim Murray]

14th, Diggy. Remember, the launch is at 8PM or something.

 

[Evildeus]

At least we will see some benchmark, for sure, on the 13th

Nope.

You are an evil, evil man! What are you talking about? Isn't the launch date the 13th, or do you think it might just be one of them paper launches with no product available.....

Well, perhaps B3D won't have some bench, but i'm pretty confident that we will see some

 

[martrox]

Well, perhaps B3D won't have some bench, but i'm pretty confident that we will see some

Hmmmm....anything like the "benches" we see at the Inq?

 

[digitalwanderer]

Well, perhaps B3D won't have some bench, but i'm pretty confident that we will see some

I'm not. Dave didn't say B3D wouldn't have them, he said he doubted we'd all be seeing them.

I tend to believe Dave, it's just a bitch to figure out what the hell he actually means!

 

[Evildeus]

Well, perhaps B3D won't have some bench, but i'm pretty confident that we will see some

Hmmmm....anything like the "benches" we see at the Inq?

I don't know (that would be ok i suppose), but well Dave says we won't Marc says we will surely so who knows

 

[martrox]

Yes....Dave really is a Vorlon.............

 

[Dave Baumann]

I don't know (that would be ok i suppose), but well Dave says we won't Marc says we will surely so who knows

Marc might need to check with Luciano.

 

[Evildeus]

I don't know (that would be ok i suppose), but well Dave says we won't Marc says we will surely so who knows

Marc might need to check with Luciano.

Perhaps he has already , but i don't know . Well, i will re-ask to clarify the thing 8)

 

[phenix]

What about the R420 benches? will they be available on 13th also?

There is a teaser at ATI's site. I think it is about R420.

We have what you want....

 

[dan2097]

2) Other things than partial precision can be used to improve performance, although NVIDIA has already promised not to use them anymore, both to the public and privately to FM AFAIK.

Are you just reffering to pixel shader optimization?

I dont get why they'd do that? they still wouldnt be able to get certifaction would they, and the public just sees that Futuremark think Nvidia cheat on the benchmark, so why not get a really big cheated score if IQ stays similar.

 

[demalion]

Pure numerology

One way to get (or, more accurately stated in this case, make up) a score of about 12500 (12557 or so):

• GT1
  280 fps
  An Athlon 64 @ 2500 MHz can get this or higher (289 for one with an FSB listing of "200 MHz").
• GT2
  108 fps
  This was based on simply doubling a 9800XT's score based on Athlon FX systems and about 500/425 MHz clock speeds.
  A typical score for a 9800XT is about 50 fps. I justified doubling based on the apparent feasibility of something like 800 MHz GDDR (well, maybe not that high, looking at what is being quoted), and high stencil and limited pixel shader functionality throughput (i.e., 16 pipelines for PS 1.3), and the apparent importance bandwidth plays in this test based on informal browsing.
  CPU and host system limitations might preclude this, but significant vertex processing throughput improvement might make it realistic.
• GT3
  94 fps
  Similar starting basis as for GT2, with the basis score for 9800XT at 500 MHz of 47 fps, and a typical 9800XT score about 42 fps.
  Same caveats as for GT2.
• GT4
  54 fps
  I stuck to a conservative theory of limited floating point throughput and approximate R300 IPC parity, and perhaps a removal of bandwidth limitations in coordination with this. Based on theories about how 16 pipelines could be managed at high clock speeds within the ambiguous transistor count guesstimates, and without exotic pipeline sharing solutions.

This fits into my current expectations for the NV40, at least by "numerology" , so I don't discount these numbers out of hand. But the Inquirer isn't known for doing fact checking, so it's still just an exercise in curiosity.

As for current fps scaling, highest fps results are about 351/67/58/58, versus about 258/51/43/43 highest stock that I noticed. There was a faster set of scores listed as stock, but the same user (and other users with system specifications within that range) had similar results with about 54?/43?, leading me to discount them as an anomoly.

 

[Sazar]

Well, perhaps B3D won't have some bench, but i'm pretty confident that we will see some

Hmmmm....anything like the "benches" we see at the Inq?

ah come on m8...

at least ONE SET of the 500 so articles and numbers and benches they have put forth so far is bound to be somewhere near the ballpark figures 8)

 

[digitalwanderer]

I don't know (that would be ok i suppose), but well Dave says we won't Marc says we will surely so who knows

Marc might need to check with Luciano.

Ok, I think I know who Marc is...but who is Luciano?

 

[Evildeus]

I suppose Nv's European PR/Marketing guy

 

[sonix666]

I can confirm that Far Cry @ 1280x1024, 4x AA runs slightly smoother on the NV40 than on the Radeon 9800XT with no AA. But hey, with 1200Mhz effective memory, what did you expect?

Uttar

Got Far Cry yesterday, am playing at 1280X1024, 4X FSAA and 8X AF, everything maxed, and it's very smooth on my 9800Pro 128 @ 430/360 on a P4 2.6 @3.4........

How do you do that? Because I have a P4 1.8A GHz overclocked at 2.2 GHz and a Radeon 9700, but the demo is FAR from smooth at 1024x768 with no FSAA and no anisotropic filtering.

 

[martrox]

I can confirm that Far Cry @ 1280x1024, 4x AA runs slightly smoother on the NV40 than on the Radeon 9800XT with no AA. But hey, with 1200Mhz effective memory, what did you expect?

Uttar

Got Far Cry yesterday, am playing at 1280X1024, 4X FSAA and 8X AF, everything maxed, and it's very smooth on my 9800Pro 128 @ 430/360 on a P4 2.6 @3.4........

How do you do that? Because I have a P4 1.8A GHz overclocked at 2.2 GHz and a Radeon 9700, but the demo is FAR from smooth at 1024x768 with no FSAA and no anisotropic filtering.

I'm running the retail game....the demo has problems with AA & AF.

 

[Hellbinder]

Ati *Dreaming* of cracking 10K...

 

[digitalwanderer]

a P4 1.8A GHz overclocked at 2.2 GHz and a Radeon 9700, but the demo is FAR from smooth at 1024x768 with no FSAA and no anisotropic filtering.

Same here with the demo, but the retail game is far smoother and it is smooth on my Barton 2500+@2.1ishGhz & 9700 Pro.