NVIDIA GeForce FX 5800 Ultra
Today we bring you a review on the heavily discussed GeForce FX 5800 Ultra. Ever since the first review came out on the web three weeks ago there has been muich talk and gossip going on. It didn't fulfill expectations the consumer had, doesn't beat Radeon 9700 Pro, uses fastest memory on the world yet uses 128-bit memory and most of all we've seen a lot of humorous images around the web where people use the FX as hair-dryer, leaf-blower. In the mean time, 3D Mark 03 was released, NVIDIA was not rather fond of this synthetic benchmark and refuses to see it as a 'gamers' benchmark for many good reasons. While we will not get into that discussion we will state this, much to my surprise nVIDIA however chose to optimize Detonator drivers for 3D Mark 03 to get the score up and actually delivers these 42.68 drivers along with the press-package I received. That by itself validates the fact that nVIDIA recognized 3D Mark 03 as common used and accepted benchmark. If nVIDIA would not decided to optimize the drivers and would not have delivered them to the press then this site would not have used the benchmark. However now we will. So there are two things in my mind regarding 3DMark 03:
- Right now there is no other way to benchmark DX9 performance .. and you need a DX9 benchmark to review DX9 hardware.
- All synthetic benchmarks are liable to be manipulated with drivers optimised for the benchmark ... this has happened since the early days of graphics benchmarking and will continue to occur. It is a moral responsibility for the graphics companies not to do this ... although there is also pressure since if they get caught cheating it will be bad for them (e.g. Quake vs Quack)
I sincerely do not see 3D Mark 03 is a definitive benchmark in the way that 3D Mark 2001 was. But we will use is it as part of our benchmark suite, we need a DX9 test, but this benchmark will a much lower rating in terms of deriving your overall score.
3DMark 03 is not a game it's a synthetic benchmark and should been seen as such, therefore out of this 20 page review and dozens of performance tests only one small page is dedicated to it. I've tried very hard to get Massive's new DirectX9 benchmark Aquamark 3 into our review suite yet:
The Aquamark 3 will be published in Q1 2003.We will inform you as soon as the Aquamark 3 is available for press and public. Please understand that we can't give away any pre-versions of the Aquamark 3 before its official publishment.
Well, I can't blame them. At least I tried people. We'll review the GeForce FX the way it's meant to be, with a lot of games. We'll do additional quality tests with AntiAliasing and Anisotropic filtering. We'll test it on an AMD Athlon and Intel Pentium 4 platform. We've also chosen not to review the card on a 3 GHz machine because ... who has such a system ?, no we'll use common used PC's in the 2 - 2.4 GHz performance range like the bigger part of you have. Also, and I must state this right away to make you aware of it, compared to other reviews we have the luxury of using more matured and better drivers. There definitely is a gain compared to other reviews on the web release a while ago. All the facts and tests will lead this review in-to a conclusion and whether you like that conclusion or not .. this is what we believe GeForce FX 5800 Ultra is all about.
You better get something to drink because this is going to be a long review. We'll start of with a piece of the technology brief as you could have read in our preview, then we undress the 5800 Ultra and show you images of her in the nude, then we'll show you some things about the drivers, then we'll test and thus benchmark it and by the time we tried to overclock the card we'll slowly move to the conclusion.
Have fun ;)