We all know that NVIDIA extremely dislikes the 3DMark03 benchmark, but it seems that up to this moment we did not know the real reason of such attitude towards the benchmark from Futuremark. I used to think that the Santa Clara, California-based developer of graphics processors tries to disregard the 3DMark03 as a reliable benchmark because its current GeForce4 Ti and GeForce4 MX products do not score high results there. However, the real reason lies even below: the forthcoming NV31 and NV34 VPUs score in 3DMark03 like ATI’s RADEON 9500 and RADEON 9000 respectively, as The Inquirer’s Fuad Abazovic reports over here.

Eventually NVIDIA may also start to vigorously fight the benchmark and prohibit the reviewers who receive graphics cards from NVIDIA to use the 3DMark03, as such low result can do a lot of difficulties for NVIDIA. At least, someone who gave a GeForce FX reference graphics card to Beyond3D web-site asked them not to use the Futuremark’s latest benchmark, as an editor points out over here. Even though such statement cannot be considered as a proof, we definitely can take it as an interesting episode in “NVIDIA vs Futuremark confrontation”, an addition to the document that NVIDIA sends out to certain journalists.

Basically speaking, NVIDIA is trying to fight a loosing battle here, just like its now dead arch-rival 3dfx did in the past. 3dfx and its fans used to explain three years ago that synthetic benchmarks have nothing to do with real games, that T&L is not needed for present and future games since DirectX 7 pipeline has so many limitations, that graphics cards from 3dfx run all games smoothly and better than NVIDIA or ATI based solutions. Some game developers even used to say that certain games run best with 3dfx graphics card what eventually became hoax. What do we see now? NVIDIA explains that 3DMark03 is not a proper benchmark, a great job for the company who distributed BenMark, Chameleon Mark and was fully satisfied with 3DWinbench and 3DMark99, 2000 and 2001 as well as with all the other synthetic benchmarks. Now some NVIDIA love bunnies explain that we do not need all those advanced DirectX 8.1 stuff found in RADEON 9000 and Xabre GPUs, as no game developers will ever use it, but stick to DirectX 9.0 features. Some game developers, including those, who, I recall, used 3dfx’s Glide as their primary platform [and failed afterwards], now put “NVIDIA: The Way It's Meant to be Played” logotype on certain titles. If this logotype really means something, I start to think that some people do not learn on their own mistakes, if the logotype does not mean anything at all, I have no idea why they put it on? Anyway, the problem is not in logotypes, but in the fact that NVIDIA, who, by the way, still sells some great graphics processors, takes 3dfx approach of competing against rivals with rather debatable statements, explanations and advertising strategies.

Even though 2389 and 839 points NV31 and NV34 manage to score in the 3DMark03 on an Athlon XP 2700+ based system with 512 MB of RAM may not reflect real-game performance now, they may represent us the situation with future 3D games. Since very few games are equipped with proper benchmark, the only way for us to get an idea about their performance is to use synthetic benchmarks, such as 3DMark03.