Nvidia is cheating with the new drivers !!!

[StanTheMan]

Nvidia is caught cheating with the new driver!!!!
geez, I'm a die hard nvidia fan, but this one makes me disappointed. go to this korean website to see the benchmark of the new driver (43.45):
http://www.darkcrow.co.kr/Review/Review_Content.asp?board_idx=146

and then see, that the dirvers that perform best in the 2003 nature games somehow does not display all the images it should be (see the light behind the tree. To see the picture comparison directly go to:
http://www.darkcrow.co.kr/image/tip&tech/4345tech/iq/game4.jpg

now i'm sure that 3dmark is actually a valid benchmark, as long as the graphic company does not produce driver that "corrupts" bits and pieces of the image quality. maybe when they release what so called "optimized driver for 3dmark 2003", they tr to test if ppl caught them cheating, but they didn't. so they finally released a final driver and incorporating the "cheat"
btw, i couldnt tried it myself as i only own g4Ti.

 

[Davegod]

nvidia publicly stated it thought 3dmark2003 a rubbish bench and wont have anything to do with it, to which futuremark responded with a monster report, nvidia release some drivers to hardware sites specifically to show how easy it is to "cheat" 3dmark2003 but are forcing many places that normally hold beta drivers not to host these ones, also futuremark are now preventing these drivers' benchmarks from being published. if they're released another set of drivers that give bogus 3dmark2003 results, it doesnt really mean much since nvidia have already discounted the benchmark publicly. the important bit imo is a poor image quality?

any benchmark is only "valid" if you use it for what it is designed for and take into account its limitations...

tbh i cant see any difference between 42.75 and 43.45 image anyway, 43.03 otoh seems to have a very different sky - but these were always intended to be public "cheat" drivers distributed to hardware sites only to prove a point.

just imho

edit: repost :/

 

[StanTheMan]

look at the reference screenshot shown in the website.
the light is supposed to be there.
and then look at the image quality of the 2 drivers that performs best (in term of speed)
the 2 drivers that didn display the light perform miracously best

btw, the 43.03 is not the drivers that cheats 3dmark. the cheater was the 42.xx driver.
if you care enought to open the website (and see the benchmark), you'll see that the 43.03 driver does not perform best in 3dmark 2003.
I think its because of the image quality manipulation in 43.45 and 42.xx driver.

btw, yea i spread the news in may 43.45 related threads, cos i was so angry abt the fact that a user are forced to sacrifice image quality due to d4mn video card manufacturer who care only to be the king of speed.

 

[XBoxLPU]

so freaking what if they are cheating..... do you play 3dmark2003 all of the time ?

I could care less


btw, and I do own an Ati 9700 Pro ....

 

[spanner]

Nvidia is trying to make a point, besides isn't this driver still unnofficial. In other words, go nvidia.

 

[JellyBaby]

When nVidia adds code to production drivers that cheats by covertly turning off settings when it detects certain games like Quake, then please let me know.

FYI in case you didn't know ATi did that and has a cheater track record dating back to the 80s.

I suspect both companies could use a bit more honesty in the driver department...I'd really like to know what that new performance slider does in the Detonator driver.

 

[ElFenix]

Originally posted by: XBoxLPU
so freaking what if they are cheating..... do you play 3dmark2003 all of the time ?

I could care less


btw, and I do own an Ati 9700 Pro ....

couldN'T



so nvidia doesn't like how the benchmark makes their hardware look so they release a driver to break the benchmark. thats really good of them. the sort of company i would do business with.

 

[AmdInside]

I am glad that NVIDIA is showing how faulty 3DMark is. For the longest of time, I have been against 3DMark as a benchmark. The only cases where I can recommend to people use 3DMark is to make sure their system is performing as it should by comparing their score to the scores of other people on the internet who have the same hardware that they do. Other than that, 3DMark is only good for eye candy.

 

[Spicedaddy]

FutureMark said that only WHQL driversshould be used with 3DMark03 from now on. I guess that solves the problem because nVidia will have a hard time getting these hack jobs certified...

 

[NFS4]

Originally posted by: AmdInside
I am glad that NVIDIA is showing how faulty 3DMark is. For the longest of time, I have been against 3DMark as a benchmark. The only cases where I can recommend to people use 3DMark is to make sure their system is performing as it should by comparing their score to the scores of other people on the internet who have the same hardware that they do. Other than that, 3DMark is only good for eye candy.

NVIDIA is "Breaking" the benchmarks with NON-WHQL drivers. That would explain this PR:

Official Statement from FutureMark Concerning WHQL Drivers

WHQL drivers wouldn't cause the misrepresented graphics shown with the 43.45's

 

[Redviffer]

Honestly, I don't see much of a difference between any of the shots. That, and all my games are still working fine with no apparent problems from the new drivers. That is what means more to me.

The last drivers fixed a bug in Unreal II (all the characters looked like chromed T2 guys except Aida - what a cutie). That's what I'm more concerned about, being able to play my games without visual defects or without crashes. As long as the drivers do that, I don't care if they please FutureMark or not.

 

[BFG10K]

When nVidia adds code to production drivers that cheats by covertly turning off settings when it detects certain games like Quake, then please let me know.

You can't because the drivers lower image quality in all games, not just one.

 

[StraightPipe]

Are the games in 3d mark released? The WW2 Fighter game (on 2003) looks sweet.


Hahha Nvidia, got busted cheating.

(cant blame 'em I've cheated before)

 

[ROJAS]

If the latest driver works for you and you are happy with your pc's performance, use it.

ROJAS

 

[Schadenfroh]

This is sorta like when ATI got caught cheating with the quake 3 benchies a while back with the 8500. At least nvidia cheated and won. ATI cheated and still lost to the geforce4. WHQL is not that big of deal, i am sure there are plenty of good drivers that are not WHQL.

 

[StanTheMan]

yea i know ati also cheats back then by trying to manipulate quake III games (I still remember quake and quack incident that time).
what concerns me is that the 43.45 is the FIRST RELEASED DRIVER that support GeforceFX line (I don't talk about leaked/beta drivers as i wouldn't touch them). the previous one (41.09) only support up to geforce 4 line. so as a (potential) user i left with no choice of running my video card with the best image quality as nvidia doesn't give us that choice with the driver.

 

[Mem]

now i'm sure that 3dmark is actually a valid benchmark, as long as the graphic company does not produce driver that "corrupts" bits and pieces of the image quality. maybe when they release what so called "optimized driver for 3dmark 2003", they tr to test if ppl caught them cheating, but they didn't. so they finally released a final driver and incorporating the "cheat"

Look at it this way how many people that install 43.45 drivers will be running 3Dmark2003,let alone all the benchmarks if their card supports DX9? I bet less then 1% of the total Nvidia owners that install them.

As for whether it`s a cheat or not,too be honest I don`t know or really care.

It amazes me why some people care so much for a benchmark,do you buy your card for gaming or benchmarking.

 

[AtomicDude512]

They should at least try to get good image quality, I mean, the ATI's benchmark high and have high image quality! Poor choice on Nvidia's part I think...

 

[BDSM]

ATI generally has superior image quality compared to NVidia.. So arguing that 3D mark is easy to cheat using low image quality actually is really stupid. Nvidia are in fact calling themselves cheaters.. DOH!

And then they bring out a driver that is made to give even poorer quality.. DOUBLE DOH!!!!

 

[human2k]

Originally posted by: XBoxLPU
so freaking what if they are cheating..... do you play 3dmark2003 all of the time ?

I could care less


btw, and I do own an Ati 9700 Pro ....

 

[chizow]

1. NVIDIA GeForce4 Ti 4200 (19671) (23%)
   
   ATI RADEON 9700/9500 Series (8446) (10%)
   
   NVIDIA GeForce4 Ti 4600 (7358) (8%)
   
   ATI RADEON 8500/LE (6648) (8%)
   
   NVIDIA GeForce2 MX/MX 400 (5735) (7%)
   
   NVIDIA GeForce3 Ti 200 (5535) (6%)
   
   NVIDIA GeForce4 MX 440 (5227) (6%)
   
   NVIDIA GeForce4 Ti 4400 (4670) (5%)
   
   NVIDIA GeForce2 GTS/Pro/Ti (4265) (5%)
   
   ATI RADEON 9000 (3395) (4%)
   
   NVIDIA GeForce3 (2328) (3%)
   
   ATI RADEON 7500 (1596) (2%)
   
   NVIDIA GeForce3 Ti 500 (1461) (2%)
   
   NVIDIA GeForce4 MX 420 (1297) (1%)
   
   ATI RADEON DDR (981) (1%)
   
   NVIDIA TNT2 m64 (927) (1%)
   
   NVIDIA GeForce2 MX 100/200 (803) (1%)
   
   ATI RADEON VE (644) (1%)
   
   NVIDIA GeForce DDR (439) (1%)
   
   NVIDIA GeForce4 MX 460 (381) (0%)

LoL, looks like FutureMarketing realized how insignificant they were in the big scheme of things. They've flip-flopped their stance on non-WHQL drivers numerous times in the last few days since their initial "stand" on non-WHQL dets. I guess they realized alienating 70% of their users wouldn't be a good idea. Btw, non-WHQL results WILL be publishable and viewable on ORB, there's just going to be some denotation whether or not the driver is WHQL certified.

I'm not too keen on WHQL certification to begin with. I see it as a mechanism that hinders progress, particularly because the fee Microsoft charges for WHQL certification and the QA process can become tedious if you are constantly making updates to drivers. I've always liked nVidia's approach in releasing non-WHQL drivers instead of waiting a few months between certified drivers, as they've always shown some improvements in games while enabling overclocking within the UDA w/out a 3rd-party tweaker.

If there's an issue in games and floating point precision, then I might be concerned, but at the end of the day, its just nVidia giving the big middle finger to FutureMark for choosing not to optimize 3DMiss2K3 for their products, again.

Chiz

 

[ZimZum]

Originally posted by: spanner
Nvidia is trying to make a point, besides isn't this driver still unnofficial. In other words, go nvidia.

So Nvidia releases drivers that lower the image quality in games and this is supposed to be a good thing?

And just to get back at futuremark? Nice to see Nvidia has its priorities in order. Customers, Schmustomers weve got to show futuremark whos boss!!

 

[Stargazer78]

Originally posted by: spanner
Nvidia is trying to make a point, besides isn't this driver still unnofficial. In other words, go nvidia.

Yep, after years of pimping on 3dmarks, all of a sudden they decide to turn against it the moment the program doesn't make their cards look as good anymore, my, go go nvidia, going down drain i reckon...