swaaye wrote:

The 4x.xx drivers do have some cheats. They were cheating with 3DMark 03 and were caught for example. If you go newer though you can run into problems with some older games.

The driver optimizations aren't all bad. I mean they were optimizing for new hardware after all. The drivers have a shader compiler and that was being worked on for ex. However with some Shader Model 2 games there are some instances that NV replaced shader code and reduced its complexity so the FX cards could render them at a reasonable rate. I remember seeing Aquanox 2 screens with visible simplications. But would you even notice without side by side comparison? It's questionable really. But yeah it's best to just avoid SM2 games on those cards. The hardware is simply not any good at SM2.

Doom 3 is an interesting one to run on FX cards. The game is quite a good example of how to get the most from them. Less math more texturing tricks. And NV has the higher stencil performance just for that sort of rendering. ATI has problems such as their hierarchical Z failing to work with the rendering due to an oversight in the functionality there so their efficiency goes down.

Just curious, which games?

appiah4 wrote:

Iris030380 wrote:

And sure, people slated the FX5800 for being slaughtered by the 9700Pro, and sure Nvidia cheated like crazy on their drivers to artificially close that gap, but the later revisions of the 5900's and 5700's FX cards were much more capable in DX9 and I think the 5950Ultra actually beats the 9800Pro (maybe not the XT though) across the board, DX8 and DX9.

That's some totally weird revisionist history shit there bro.

It's not secret that Nvidia released cheating drivers for their FX series ONLY because of their relatively poor performance compared to the 9600, 9700 and 9800 cards lead. Everyone knows that.

And the 5700 cards which came later in the FX series did have much better performance than the 5600 Ultras and 5800's, as did the FX5950's. Too late, but still some improvements were made.

Again, I thought that was common knowledge, bro...

The Serpent Rider wrote:

They didn't. […]
Show full quote

later revisions of the 5900's and 5700's FX cards were much more capable in DX9

They didn't.

Yes they did.

which made the Ti4600 look like an older gen card when really it was still a beast, as long as you avoided DX9.

There are plenty of DX8 games which GF4 Ti can't handle well due to heavy shader usage.

Sure, but hardly any around the time of the Ti4600 / Ti4800SE's reign. Those cards were blisteringly fast in their time. DX 8 titles were still released years later, when people had moved well beyond the GF4 series.

I think the 5950Ultra actually beats the 9800Pro (maybe not the XT though) across the board, DX8 and DX9.

Nonsense.

Many reviews will prove otherwise. Go take a looksee.

Sure, but hardly any around the time of the Ti4600 / Ti4800SE's reign.

No, most of those those games were released around 2003-2004, when GF4 Ti cards were still relevant.

Many reviews will prove otherwise.

They didn't. Games with Shader Model 2.0 had horrible performance on any FX series card and practically all reviews documented that.

mothergoose729 wrote:

Just curious, which games?

The only one I remember offhand is Aquanox 2. I doubt it's even known how many games have shaders that are modified while running by the NV drivers. Or ATI drivers for that matter. And renaming the executable to test for speed changes won't necessarily work. They don't always use a simple filename check.

Might be able to look at driver release notes and see what games they bragged about big speed boosts with.