Futuremark Responds to Nvidia 3DMark03 Criticism

[Pariah]

If it's there to normalize each test so that they approx. account for 25% each, then the DX9 test (GT4) isn't more important than the other 3 test, which Chizow implied. (exact words: I'm sure you know that the actual DX 9 capabilities are limited to 2 of 10 tests in GT 4, yet that test carries the heaviest weight.

The test that contains DirectX9 features counts no more than the other 3 tests, but what Chizow is saying is still valid. The DX9 features in game 4 are only 2 out of 10 parts. This means if you are missing one feature, which constitutes 20% of game 4 and only 5% of the total benchmark, you automatically get a 0 for 25% of the entire benchmark. That makes that feature worth much more than any other single feature in the benchmark.

You didn't explain why a couple of cards are winning game 2 and 3 by so much despite only being DX8. DX8 is current technology, so after you do, I'd like to seem games that mirror 3DMark's results.

 

[Markfw]

Question, I just got a Radeon 9500 and upgrded to DX9, and already had Cat 3.1 drivers. Why does test 4 show as "unsupported" ??

 

[Nemesis77]

Originally posted by: chizow
I don't have the time or the energy to point out all the inaccuracies in your replies, I'll just leave you with this thought:

I find your posts to be misguided, biased and naive. But I'm too lazy to point them out . So HAH!

Wait until Doom 3 is released (bout a month or so), since it seems to be the most common (and only) argument for the validity of 3DMark2K3 as a valid benchmarking tool.

So because of ONE game, 03 wont be valid benchmark? Yeah right. And like I said, there are other things besides shaders when it comes to performance.

I think you'll find that the ATI family cards (R300 and above, since the older R200 and R250 parts just suck no matter which way you look at it) will not see a 3x to 4x advantage over a "crippled" part that lack DX 9 or DX 8.1 features used in 3DMark2K3.

I guess we will find out. And even if there is no 3x difference that does not make 3DMark invalid benchmark IMO

Btw, I don't need a crystal ball to realize the majority of currently released games are as CPU dependent as GPU dependent.

Again you are talking about current games, when the point of 3DMark is about games of the future. And the benchmark is dependant on the vid-card by design.

 

[Nemesis77]

Originally posted by: Pariah
For some reason the 9500/9700 series ATi cards kill the competition in tests 2 and 3 as well which makes one wonder how that can be, when DirectX 8 games are not future tech, they are out now with completely different results. 3DMark2001SE is a DirectX 8.1a benchmark, why doesn't it give the same odd results that 3DMark 2003 does in games 2 and 3? Which benchmark is the better example of real game preformance?

Aquanox

"Aquanox makes heavy use of pixel and vertex shaders in the DirectX 8 standard."

Does Aquanox use PS1.1 or PS1.4? It was designed with GF3 in mind, so I would assume it uses PS1.1 at most. And in that benchmark, I see GF4 losing. and you claim that 03 is a biased benchmark, yet you use Aquanox as a counterexample. Considering that:

"Massive Development has been working closely with NVIDIA to expose the awesome power of GeForce3 in their forthcoming title AquaNox," said Sanford Russell, Senior Director of Partner Management at NVIDIA. "Geforce3's nfiniteFX? engine has enabled Massive Development to bring real-time Hollywood quality to their game. Only by using NVIDIA's hardware Vertex Shaders and Pixel Shaders, are they able to achieve their unique lighting solution at great frame rates. AquaNox and its technology, the krass?engine impressed us! NVIDIA looks forward to working with Massive Development again in the future to take games and gamers to the next level."

(Source)

Is Aquanox unbiased benchmark?

And besides, this isn't about GF GX, but about NV's older lineup.

Again, similar situation, much different results. You want to claim that 2003 is for future games, but even the DirectX 8 benchmarks give bogus results when compared to current games that use the same features.

DX8 = PS1.1. DX8.1 = PS1.4. There is a huge difference between the two. Since Aquanox doesn't use PS1.4, Ati has no advantage when it comes to shaders.

They might as well slap a "Built By ATI" sticker on it and call it "The ATI Video Card Benchmark". Its certainly not "The Gamer's Benchmark", and does nothing to mirror real-world gaming performance, in future or current games.

Why? Because it uses tech that is not supported by NV's crippled vid-cards? Why do people insist that benchmarks must be designed for NV's feature-set? Only then the benchmark would be "unbiased"

?

 

[Nemesis77]

Originally posted by: NOX
I?m not just talking about Nvidia or HardOCP. My point is this has been discussed over and over and over again on these forums. People, knowledgeable people for that matter, have stated over and over again, that 3dmark0x is not a suitable benchmark to measure REAL WORLD PERFORMANCE.

NV and HardOCP are among others saying that. If that is the case, why did they happily use 3DMArk01? Why use a benchmark that is not a suitable benchmark to measure "Real world performance"? Why did they use it? Obviously both HardOCP and NV thought that 3DMark is a suitable benchmark, now it suddenly is not.

 

[MelvinTong]

Hi Guys,

I hope I can get some questions/comments from the public as well.

I have contacted NVIDIA and they will respond to some of the questions you all have. Please tell me the questions you want to ask NVIDIA and I will send the questions to them to be answered.

Please make the questions clear. The questions should be related to Futuremark's 3DMark03 (or their optimized drivers). Please get me the questions ASAP. Thanks!

 

[Nemesis77]

Originally posted by: MelvinTong
Hi Guys,

I hope I can get some questions/comments from the public as well.

I have contacted NVIDIA and they will respond to some of the questions you all have. Please tell me the questions you want to ask NVIDIA and I will send the questions to them to be answered.

Please make the questions clear. The questions should be related to Futuremark's 3DMark03 (or their optimized drivers). Please get me the questions ASAP. Thanks!

1. Why does NV still offer DX7 vidcards in their mainstream whereas your competitors have already moved to DX9?

2. Would NV have any problems with 3Dmark03 if all your cards were DX9-compliant, instead of being still limited to DX7 (The MX-series)?

3. Why didn't NV implement PS1.4, since it's clearly superior to PS1.1?

4. Why should Futuremark design their benchmark with just NV's featureset in mind?

5. If 3DMark is not a valid benchmark and you should use real games instead, why did NV use 3DMark01?

 

[MelvinTong]

Originally posted by: Nemesis77

Originally posted by: MelvinTong
Hi Guys,

I hope I can get some questions/comments from the public as well.

I have contacted NVIDIA and they will respond to some of the questions you all have. Please tell me the questions you want to ask NVIDIA and I will send the questions to them to be answered.

Please make the questions clear. The questions should be related to Futuremark's 3DMark03 (or their optimized drivers). Please get me the questions ASAP. Thanks!

1. Why does NV still offer DX7 vidcards in their mainstream whereas your competitors have already moved to DX9?

2. Would NV have any problems with 3Dmark03 if all your cards were DX9-compliant, instead of being still limited to DX7 (The MX-series)?

3. Why didn't NV implement PS1.4, since it's clearly superior to PS1.1?

4. Why should Futuremark design their benchmark with just NV's featureset in mind?

5. If 3DMark is not a valid benchmark and you should use real games instead, why did NV use 3DMark01?

Hmm, thanks for your questions Nemesis. But before that, I do not want to be questioned by NVIDIA if I submit them my questions. So I need more 'facts' to support the questions.

1. NVIDIA uses DX7 on their lower end cards for the the value and 'acceptable' performance. There are still a number of softwares running on DX7.
2. PS 1.4 is implemented in GF4 (correct me if I am wrong) and PS1.1 is in GF3.
3. NVIDIA claims 3DMark03 is not a valid benchmark, not 3DMark01.

 

[Priit]

IMO, Nvidia is partly right, but previous versions of 3DMark has been just as meaning(less/ful) as the new one and they have hyped those neverheless. Point is that there are so many different differently performing games out there it's just impossible to make conclusions about their performance with just one limited functionality demo (what 3DMark IMHO really is). BTW, 3DMark99 MAX showed K6-2 to be much more powerful/faster gaming processor than P2, too

 

[Nemesis77]

Originally posted by: MelvinTong
Hmm, thanks for your questions Nemesis. But before that, I do not want to be questioned by NVIDIA if I submit them my questions. So I need more 'facts' to support the questions.

1. NVIDIA uses DX7 on their lower end cards for the the value and 'acceptable' performance. There are still a number of softwares running on DX7.

Ati has "value" cards that are full DX9-compliant. Only 9100 is not DX9-compliant

2. PS 1.4 is implemented in GF4 (correct me if I am wrong) and PS1.1 is in GF3.

Negative, GF4 support PS1.1 (not sure about 1.2 and 1.3, but they are minor improvements so they don't matter)

3. NVIDIA claims 3DMark03 is not a valid benchmark, not 3DMark01.

Why is 01 a valid benchmark where 03 is not?

 

[Turkey22]

my friend has and 1800+ and a Ge4 4200 and it almost killed his machine. Almost made him cry cus he upgraded recently.