3dmark rigged?

[taylor34]

Hey Guys--

Saw these two threads over at rage3d and nvnews about how disabling the splash screen causes geforce 4 cards to take a 600 to 1000 point drop in 3dmark 2001 (ati cards unaffected). Pretty interesting anyways--here are the links:

http://www.rage3d.com/board/showthread.php?s=&threadid=33621345

http://www.nvnews.net/forum/showthread.php?s=&threadid=15459

Can anyone else verify this?

Taylor34

 

[Acanthus]

Ive heard nothing of this, i didnt even know you could disable the splash screen. To my knowledge 3DM2k1 has a LOT of serious cheat proof code in the software, although it has been done (the famous Dell) its very hard to cheat on 3dm.

 

[taylor34]

http://www.theinquirer.net/12060214.htm

Here's a link about the story--nvidia says it's a bug, lol. Isn't that the same thing that ATI said about the quake/quack thing? I'm not sure why this isn't a bigger story--when ati did this it was like the end of the world. When nvidia does the same thing, no problems whatsoever. And to anyone who honestly thinks that this is a bug--how is that possible? If I told you quake ran 10% faster if I enabled a splash screen, what would you say? What does a splash screen do in a benchmark, except to give you a way to detect when a certain benchmark is being run. Just my opinion though--there's no way a low-level driver should be affected by splash screen.

Taylor34

 

[Gonad the Barbarian]

Is it not possible that the bug is somehow costing performance when the spash screens are disabled? Everyone seems to be assuming that the higher scores with the screens enabled are what need to be 'fixed' and not the lower scores with the screens disabled.

 

[nemesismk2]

IMO 3Dmark 2001 IS rigged, it doesn't even reflect real world performance at all. Let's all stick 3Dmark where it belongs, right up Madonion's a*s because they talk out of it most of the time anyway (especially that worm person)

Madonion recently made a huge fuss about a program called 3D Analyser which "cheats" but their own program is probably even worse, they say it's a reliable benchmark, yeah that might be true but only if your video card is made by Nvidia.

You want proof here it is, Madonion/Nvidia got the problem with the splash screen fixed within a day but when it was found that 3Dmark2001 had problems with Kyro2 video cards it was a YEAR later before it was fixed by Madonion.

 

[merlocka]

You want proof here it is, Madonion/Nvidia got the problem with the splash screen fixed within a day but when it was found that 3Dmark2001 had problems with Kyro2 video cards it was a YEAR later before it was fixed by Madonion.

Not quite. This has been present since the Geforce 3. It's in Madonion's FAQ.

Madonion recently made a huge fuss about a program called 3D Analyser which "cheats" but their own program is probably even worse, they say it's a reliable benchmark, yeah that might be true but only if your video card is made by Nvidia.

Can you point out some popular game benchmarks in either openGL or D3D which consistantly contradict Madonion's results?

Top 15 from MadOnion

NVIDIA GeForce4 Ti 4600
NVIDIA GeForce4 Ti 4400
NVIDIA GeForce4 Ti 4200
ATI Radeon 8500
NVIDIA GeForce3 Ti 500
NVIDIA GeForce3
NVIDIA GeForce3 Ti 200
NVIDIA GeForce4 MX 460
NVIDIA GeForce4 MX 440
NVIDIA GeForce2 Ultra
ATI Radeon 7500
NVIDIA GeForce2 GTS/Pro/Ti
NVIDIA GeForce4 MX 420
ATI RADEON DDR
NVIDIA GeForce DDR

Top 15 from Digit-Life (note, I modified the list to "clump" smiliar cards together. actual ranking is here)

Digit-Life uses Return to Castle Wolfenstein, Quake3 QUAVER, Comanche4, Giants, 3dmark2001, Serious Sam, and iXBT RightMark Video Analyzer in their benchmarks.

NVIDIA GeForce4 Ti 4600
NVIDIA GeForce4 Ti 4400
NVIDIA GeForce4 Ti 4200
ATI RADEON 8500
NVIDIA GeForce3 Ti 500
xxx Geforce 3 not tested xxx
NVIDIA GeForce3 Ti200
NVIDIA GeForce4 MX 460
ATI RADEON 7500 64 MBytes
NVIDIA GeForce4 MX 440
NVIDIA GeForce2 Ti 200
ATI RADEON 64 MBytes
NVIDIA GeForce4 MX 420
NVIDIA GeForce2 Pro 32
ATI RADEON 32 MBytes DDR

So basically, Digit-Life shows the Radeon 7500 performing a bit better. But in general they both reflect the results which most review sites obtain, among a variety of both D3D and OGL benchmarks. So where's the problem with Madonion?

In addition, in Anand's Unreal 2003 test the cards stack up like :

NVIDIA GeForce4 Ti 4600
NVIDIA GeForce4 Ti 4400
NVIDIA GeForce4 240/500
NVIDIA GeForce3 Ti 500
NVIDIA GeForce3
ATI Radeon 8500
ATI Radeon 8500LE
NVIDIA GeForce3 Ti 200
NVIDIA GeForce2 Ti 200
NVIDIA GeForce4 MX 460
NVIDIA GeForce4 MX 440

Radeon's get bumped a notch down, but the trend is of course the same.

I agree that many people who sit around and tweak their computer to have the highest possible 3dmark score is kinda silly, but it seems to me that the bechmark is quite realistic in ranking Direct3D performance for 2001/2002 games.

 

[Crapgame]

I'm still wondering what the big problem is? So the splash screens give the GeForce's a second to precache the next test, sound perfectly logical to me.

 

[HowDoesItWork]

I think what people are saying is that it is supposed to be a real world type test. Since most of us don't see a splash screen every 30 seconds during real world type use, having that splash screen be a 'pre-cache' time period is not a realistic test.

Having said all that, I agree with Merlocka. 3dmark might not be perfect, but it seems to give you a generally accurate picture of how hardware compares.