Nvidia/ATI cheating on 3D benchmarks!

[meerschaum]

http://www.extremetech.com/article2/0,3973,1115365,00.asp

the scandel continues... I dunno how many of you guys have followed it...but its been a constant drama for a couple weeks now...

[Hammer]

Well, it's not a cheat. GPU/VPU is like a CPU in regards to applying optimisations.

[DethKnight]

I dunno how many of you guys have followed it...

I followed it since ars and the reg had it, but now I've lost interest init

benchmarks are made to be fudged, I'd rather rely on my own

RWO's

[iamaboringperson]

i havnt read the article yet, but fudging up benchmarks is nothing new - sometimes they just optimise the software, cpu manufactures use custom compilers

[kzin]

All I know is my Radeon 9700pro kicks ass in unreal 2 who cares about bench marks

[mbilla]

Hi

@meerschaum

the scandel continues

why scandal?

As far as I know all benchmarks made by companies are faked. Benchmarks don't bring anything.
Try a card if it fits your needs it's ok, if not try another one.

Or what about car manufacturers? The fuel usage they advertise never matches the reality.

Life is very easy: politicians, laywers and managers are all of the same kind: they are there to take your money in exchange for fake products or informations.

I never saw a a politician who did what he /she promised in his / her electaion program.
Laywers do live at the border of good and evil and they take money from both sides.
Managers do sell products which are all TOO expensive (they earn at least 30% on any product) and NEVER can do all the things advertised.

So why worry about benchmarks?

[mikeymike]

A graphics benchmark is supposed to be a definitive indicator of graphics card performance.  Optimising for a graphics benchmark does not 'enhance the user experience' in any way, it is only cheating users, that is the whole intention of the exercise.  It is a cheat.  Optimising for games is different as that IS 'enhancing the user experience', because they want decent games performance.  How many people do you know who would love a graphics card that only performed well in benchmark tests?

Furthermore, ATI and NVIDIA are the two biggest graphics card companies.  Once the benchmark designer people cave in and allow them to 'optimise' for their software, they're shutting out the smaller/new competitors whose time is more critical.

It's like a teacher saying to a parent "your son is great at maths! Sometimes he has a few problems remembering the answers we give him to the exam questions, but after some prompting, he got 100% correct! Isn't that great?".

[Hammer]

mikeymike wrote:
A graphics benchmark is supposed to be a definitive indicator of graphics card performance.  

The ideal benchmark should be composing of the leading 3D engines e.g.
UT, UT Warfare Engine, UT2003, DOOM 3, QuakeII, QuakeIII, Westwood Studios (e.g. C&C General), Aqua Mark, Relic (e.g. Homeworld), and ‘etc’ (i.e. Top 10 3D game engines).

FutureMark’s 3D engine is not as popular as the other 3D engines in the market thus throwing FutureMark’s claims of “gamers benchmark” into doubt…  

[B00tDisk]

I think I don't give a damn; after about 45fps at 4x FSAA 1280x1024 the eyes just quit registering a difference...

[jdiffend]

Companies have been cheating benchmarks for over a decade... why stop now?