Splinter Cell: there might be another.

[Typedef Enum]

http://www.beyond3d.com/forum/viewtopic.php?t=6038

Beyond3D is currently looking into the Splinter Cell benchmark that they (B3D) created.

To cut to the chase, there's a sequence where a Pixel shader is supposed to be applied to the ocean surface...and interestingly enough, this doesn't seem to be happening all the time with the 5900.

Furthermore, the thing that's fueling the speculation at this point...beyond the fact that we now know that nVidia is a proven benchmark cheater...is the fact that they have been tipped off that an nVidia engineer's main mission in life was to focus on Splinter Cell benchmarking.

I strongly believe this is only the beginning.

[DSC]

Please ask DaveB and Rev to take screenshots, like the one DaveB did for GT4. This will help show clearly whats happening.

[Skinner]

Well, with driver 43.51 there are also missing some sun/flares /beams sometimes, but I don't know if its a cheat.

[jAkUp]

hmmm... this might just be a driver bug since he did say that the water effect appeared on the 5200

[Typedef Enum]

It could very well be that such an "optimization" was pointless on the 5200.

[StealthHawk]

I think it's premature to try to bring this to light. Since it occurs only on the gfFX5900 and gfFX5600, but not the gfFX5200, it might be cheating.

But B3D is not ready to come to such a conclusion without more testing. So you shouldn't jump to conclusions yet either.

Although it is interesting that Splinter Cell benchmarks jumped up with recent drivers, as did Shadermark results...I honestly wouldn't be surprised if nvidia was cheating.

P.S. I haven't read the thread today, just last night, so there may be new developments I missed. If so, just disregard my comments.

[PreservedSwine]

Even thoughnVidia cheated in 3Dm03, I'll give them the benefit of the doubt on yet another cheat. Would this bug/cheat affect the benchmark, or the game as well?

I'll reserve any judgment till all the facts are in....

[Dazz]

Last i saw was that the GeForce FX range didn't render any special effects in Splinter cell, even the GeForce 4 Range rendered them. But perhaps thats been fixed but i havn't heard anyone say that it has? It's funny really as i though the game was based around the GeForce FX.

[ChrisRay]

The shaders being rendered are DirectX 8 Shaders, Which is nothing remarkable.The Geforce FX has no problem with 1.1 Pixel Shaders.

I think they have jumped the gun on this. Where a bug is present. I really doubt this is a cheat.


The geforce FX handles DirectX 8.0 shaders fine. Heck Nvidia helped microsoft design the DX 8.0 Shader standards


In Regards to them focusing on this engine. I believe its important. As this engine is performing rather underwhelming on both ATI AND Nvidia cards right now

[Dazz]

I wouldn't really say that as frame rates don't need to be as high for this game. Also saying cards an't perfoming verywell is an understatement, the Radeon 9600Pro is considerbly faster then a Radeon 9500Pro.

[ChrisRay]

Originally posted by Dazz
I wouldn't really say that as frame rates don't need to be as high for this game. Also saying cards an't perfoming verywell is an understatement, the Radeon 9600Pro is considerbly faster then a Radeon 9500Pro.


I'm still trying to figure the 9600 thing out with this game. I can only assume this game benefits from higher bandwith rather than higher fill rate, And this game is not pushing fill rate at all.

Since you can't do any Anti Aliasing in this game currently. There's no real way to test that theory

[DaveBaumann]

Originally posted by ChrisRay
The shaders being rendered are DirectX 8 Shaders, Which is nothing remarkable.The Geforce FX has no problem with 1.1 Pixel Shaders.

I think they have jumped the gun on this. Where a bug is present. I really doubt this is a cheat.

The geforce FX handles DirectX 8.0 shaders fine. Heck Nvidia helped microsoft design the DX 8.0 Shader standards

We've not actually stated anything, other than it looks odd. And, yes, you are correct - the FX should be fine with PS1.1, which is why I find it odd that this issue should occur in the first place. Its even odder that it occurs on some parts of the same architecture but not others.

Originally posted by ChrisRay
I'm still trying to figure the 9600 thing out with this game. I can only assume this game benefits from higher bandwith rather than higher fill rate, And this game is not pushing fill rate at all.

I thinks its evident that the R300 ASIC that 9500 uses had some issues with PS1.1:

http://www.beyond3d.com/reviews/ati/r350/index.php?p=15

Both R350 and RV350 cleaned these issues up, and the difference in performance with SC is just a real world example of that.

[ChrisRay]

Originally posted by DaveBaumann
We've not actually stated anything, other than it looks odd. And, yes, you are correct - the FX should be fine with PS1.1, which is why I find it odd that this issue should occur in the first place. Its even odder that it occurs on some parts of the same architecture but not others.



I thinks its evident that the R300 ASIC that 9500 uses had some issues with PS1.1:

http://www.beyond3d.com/reviews/ati/r350/index.php?p=15

Both R350 and RV350 cleaned these issues up, and the difference in performance with SC is just a real world example of that.


Well, I think its possible that both companies have not properly optimised for this game. I do really believe it to be strange that the FX is having issues with with 1.1 shaders,

But we probably should note that the FX 5200 architecture is a bit different than the 5600 Ultra and 5800 ultra, Due to the z compression ect,

So its not improbable to assume it could be related to that. If Nvidia "is" cheating in the benchmark. The Results are very under whelming.


Interesting note, About the r300 and Splinter Cell. perhaps there's hope for performance in this game yet

[fantomas]

hmm, i wanted to say this in a different thread, but it is closed due to flames...

Who cares? Do you really think ATi will have a better card when Doom III comes out? I doubt it. So the card isnt quite as good as what the benchmarks are saying, so what? Do you think nv35 will be worse than your MX?

My point is: I would happily put an nv35 in my system :) If you have one and hate it or nvidia send it to me, i will game on with this cheater :D

[jjjayb]

Do you really think ATi will have a better card when Doom III comes out?

Count on it.

As far as cheating on splinter Cell. I think it's too early to think that.

[Typedef Enum]

I personally think none of the current ones, be it the 9800 Pro, or the NV35, will be the ideal card to play DoomIII...Yeah, the game that we can only _hope_ will be released by the end of the year.

If you were to give ATI a good heads up on going one-on-one with nVidia in Doom III benchmarks, I have very little doubt that the 9800 Pro would either be on par, or even slightly ahead...

[Hanners]

Originally posted by fantomas
Who cares? Do you really think ATi will have a better card when Doom III comes out? I doubt it.

Yes, they will. In fact, they may even have two better cards out by the time Doom III is released.

[Morrow]

Originally posted by Hanners
Yes, they will. In fact, they may even have two better cards out by the time Doom III is released.

Two? The R360 as shown by TheInquirer is just again an overclocked 9800 Pro (like the 9800 itself was). No new features, no 0.13 micron manufacturing process.

The R420 will be available in stores January 2004 the earliest! So, where are those two new fast cards coming out before Doom3 is released?

[StealthHawk]

Originally posted by Morrow
Two? The R360 as shown by TheInquirer is just again an overclocked 9800 Pro (like the 9800 itself was). No new features, no 0.13 micron manufacturing process.

The R420 will be available in stores January 2004 the earliest! So, where are those two new fast cards coming out before Doom3 is released?

He was implying that Doom3 might be delayed...again.

[ChrisRay]

Originally posted by StealthHawk
He was implying that Doom3 might be delayed...again.


Seriously? is this due to the Microsoft/Xbox crap?

[StealthHawk]

Originally posted by ChrisRay
Seriously? is this due to the Microsoft/Xbox crap?

Who knows. When is Doom3 actually slated to come out?

I have always believed it would be sometime this year. Carmack said he wanted the game to be finished around now. But obviously it isn't finished. The version tested in the NV35 previews is still 0.02. That doesn't instill me with confidence that it will be ready soon.

Meanwhile, Valve has said that Half-Life2 will be out before Doom3. HL2 comes out September 30 if everything goes as planned. We still don't have any word AFAIK on when Doom3 is scheduled to come out.

[ChrisRay]

Originally posted by StealthHawk
Who knows. When is Doom3 actually slated to come out?

I have always believed it would be sometime this year. Carmack said he wanted the game to be finished around now. But obviously it isn't finished. The version tested in the NV35 previews is still 0.02. That doesn't instill me with confidence that it will be ready soon.

Meanwhile, Valve has said that Half-Life2 will be out before Doom3. HL2 comes out September 30 if everything goes as planned. We still don't have any word AFAIK on when Doom3 is scheduled to come out.


I guess the delays are too be expected. But I must admit this is quite discouraging. I am very interested in playing this game for various reasons. Mainly cuz I want to see how well my computer going to hold up against the newer engines.

As I have stated anymore. I won't be buying a new graphic cards till the r400/nv40 refresh products arrive ;\