In a salding article, digit-life.com accuse both ATI and nVidia of continued tampering of benchmarking results and even of deceitful treatment of the consumer. Following the debate sparked by the Futuremark "cheating" announcement, digit-life carried out their own investigation into what drivers actually do in the background during tests. What is worrying is that all the drivers tested were released after Futuremark's patch 330 which claimed to have fixed the reliability problem.
Digit-life even go as far as to suggest that Futuremark's part in all this controversy is not that innocent. The editor asks the company how it is possible for an amateur to discover these modifications and remove them Weren't they (Futuremark) qualified enough to locate and do away with the cheats? he adds.
The way those cheats work is, according to Digit-life.com, by aggressive optimization of anisotropic filtering and by shader code modification. The drivers know when to apply these tricks by a sophisticated application detection system which in nVidia's case was reported to cover about 70 elements.
The article also questions the motives behind nVidia's performance boosting drivers and suggests that these are not gameplay improving modifications which happen to, favorably, distort benchmark results since all optimizations detected and locked work in only one benchmark, which indicates that it took a lot of time for the programmers to analyze its behavior and specific character and bottlenecks of certain tests...
Using the Detonator 44.03 drivers and a GeForce 4 Ti 4600 card the editor managed a decent 12457 score on 3D Mark2001 SE while with the same setup but also using his own NVAntidetector scripts he dropped to just 11014. The tester justified the use of 3DMark2001 SE by claiming that it would disprove NVIDIA's statements made right after the release of patch 330 which claimes that the 3DMark2003 tests didn't comply to what we could see in future games. Digit-life also suggest that 3DMark2001 SE is more relevant to todays games than the current version.
The article, a link to which you can find by following the download tab above, also offers the necessary files so that anyone interested enough can try and repeat those results.
ATI
The more surprising aspect of the Digit-life investigation is that it presents, apparently, compelling evidence that ATI are no strangers to benchmark tampering as well.
Using the RADEON 9500 64MB video adapter with the RADEON 9500PRO capabilities unlocked with the Catalyst 3.4 drivers and 3DMark2001 SE the researcher received an 11636 score while with the same parameters but while running an ATIAntiDetector script the score was 11291.
So it is suggested that ATI are also involved in benchmark cheating. The article also claims that ATI remove only those tampering methods which are found-out while maintaining other such mechanisms.
Could we explain the Futuremark announcement and subsequent patch release as an attempt to check nVidia and bring its tampering down to ATI levels? What difference does it make to the consumer if both companies are using cheating drivers? The answer to both questions is that the consumer is treated as a naive moron and fed a variety of numbers designed to create a need and desire for an overpriced product, a need which simply does not exist.
It is important however, not to pass judgement quite yet and await the response of the vendors. A response which should come sooner rather than later in order to demonstrate the respect those companies have for their customers. We certainly do not share the researchers view that the next driver versions will try to prove that the (tampering disabling) scripts work incorrectly and disable a vital part of the code.
