NVIDIA benchmark cheats?

  somar 09:30 05 Jul 03

Two things, where can I get cut down version of Windows 98SE (less than 40MB in size) And second, did any one of you heard anything about those Russian guys who analysed NVIDIA and ATI drivers and found how they are cheating in different benchmarks?

  leo49 09:46 05 Jul 03

1 - search the forum archive for Hyperos and Strewth to find the in depth threads by flecc

2 - Don't recall the Russian connection,but I did read about this in last month's mags - didn't pay an awful lot of attention 'cos aren't benchmarks only of importance to Anoraks?

  Sir Radfordin 09:47 05 Jul 03

click here= for lots of links to stuff about Nvidia cheating

  AMD 4 ever 11:10 05 Jul 03

this is not really cheating is it? surely they are writing software that produces the best results for their hardware!

  somar 17:17 05 Jul 03

From what I understand they designed their Detonator drivers so that they can recognise application (like some specific benchmark) and than drivers will enable / disable some functions of the drivers / card and by doing that will speed up, but for some unknown application, like for example upcoming Loc On flight sim, they will use standard set up and frame rate will drop down dramatically. (If my interpretation is wrong, please correct me)

  Erlich 08:53 07 Jul 03

Please visit , click here

  Jester2K II 10:21 07 Jul 03

somar - you are correct. I understood it that the drivers would detect the benchmark programs and drop the picture quality to raise frame rates. This would give an inflated "score" making the cards appear to run faster than they do....

  DieSse 10:35 07 Jul 03

But when is cheating cheating??, and when is it optimisation?? - after all, presumably all gamers want optimised drivers, due their obsession with frame rates?

I have no axe to grind either way - I just raise the question.

  Jester2K II 14:25 07 Jul 03

Its cheating when you set the benchmark to run on two systems running identical set ups but different cards. The drivers in card A detect they are being used in a benchmark and drop the quality slightly to edge ahead of card B which doesn't "cheat". This is done knowing that people place great importance on benchmark results to compare products. Also these "optimisations" are not used in the real world when running a game so why in the benchmark if not for cheating?

Imagine you are after buying a car and you want the fastest. Car A is faster than Car B in tests. Later you find out car A doesn't perform quite as it did in the tests. The reason? When it was tested loads of unneeded weight was removed (spare tyre, jack, rear seats, dashboard etc etc) so it had a slight weight advantage over car B. Cheating or Optimisation?

  tenaka 16:49 07 Jul 03

i prefer the term choptimisation. or optimeating.

on a serious note, din ATI get caught doing a similar thing a while back, and didn't nvidia try to make a big deal about it?

  ordep 07:31 08 Jul 03

Bit more on the subject. click here

This thread is now locked and can not be replied to.

Hands-on: Samsung Galaxy S8 review

1995-2015: How technology has changed the world in 20 years

Samsung's beautifully designed Galaxy S8 makes for better VR experiences too

47 iPhone camera tips to help you take better photos