The new drivers

[Ratters]

Has anyone seen a decent comparsion on the net for the 44.65 drivers using a Geforce FX .. i will be intrested to see if there are any performance increases as this might sway me to getting one ...

[Riff]

Moving from 44.03 to 44.65 raised my UT2003 flyby score from 102 fps to 153 fps at 1024x768, 4xAA, 8xAF, high quality settings. The botmatch score went from 53 to 62. This is on an xp2800+, but the drivers seem to be pretty slick.

[Ratters]

Intresting i would like to see more reviews done based no these if they are rasing performance without losing IQ.

[Ratters]

Whats you FX ???

[jAkUp]

well my 3mark03 score jumped about 1,500 points with the new drivers.

[Paul]

Here we go again.

[bkswaney]

Originally posted by Paul
Here we go again.


No the new drivers do not act anything like the cheating drivers.
I have a PC here with a 9500 pro I'm working on and I've run it on both PC's. Mine and his.
They look the same on IQ. :)

Maybe Nvidia has learned that they will up a creek if they cheat.

[ChrisW]

I'm sorry, but just how can you automatically accept that nVidia is not cheating in these drivers and magically increase performance in benchmarks? They have never stated they would remove any of these cheats. If they had to cheat to get their high scores in the last drivers then how do you think they are getting even higher scores without cheating? Man are you guys gullible!

[Megatron]

I agree with you ChrisW.
Nothing is free. For someone to grab 1500 points in 3DMark on a simple driver upgrade?? Lol...something is definitely going on in that driver....

[surfhurleydude]

These things are THE BOMB for Unreal Tournament and Warcraft III at least...

Warcraft III ran like **** on my GF FX 5600 Go with 43.36 WHQL drivers... put these bitches on and BOOM WCIII is good to go. Gained over 35 FPS in UT2003 Flyby and 5 FPS in UT Botmatch without AA or Ansio.

[bkswaney]

Originally posted by Megatron
I agree with you ChrisW.
Nothing is free. For someone to grab 1500 points in 3DMark on a simple driver upgrade?? Lol...something is definitely going on in that driver....

U say that but I guess u forget that ATI got a 3000+ performance jump back with the 8500 with new drivers.
U can do it.

[jAkUp]

its not 1500 above the already cheating scores... its like this

3dmark03 realeases 330 patch.. scores suck
nvidia releases new drivers, scores are back up even with 330 patch

whether or not this means nvidia is not cheating, or they found a way to get around the 330 patch, i dunno. all i know is the scores are back up to the way they were before the patch.

[ChrisW]

Futuremark stated all they did was reorder some of the registers and nothing else. That was enough for the old drivers to fail to detect the shaders in 3DMark. All nVidia had to do was detect the slightly different shaders with the new drivers. I'm sure they added code to detect if the end user has moved "off the rail" and turn off any cheats so they would not be visible.

[StealthHawk]

Originally posted by ChrisW
Futuremark stated all they did was reorder some of the registers and nothing else. That was enough for the old drivers to fail to detect the shaders in 3DMark. All nVidia had to do was detect the slightly different shaders with the new drivers. I'm sure they added code to detect if the end user has moved "off the rail" and turn off any cheats so they would not be visible.

Which doesn't really explain how people have reported that the image quality is no longer decreased when taking a screenshot of shaded stuff and zooming in on it. Can nvidia really detect when a screenshot is being taken and re-render it? Maybe I'm missing something.

OTOH, it is awfully fishy that the score with new drivers is exactly the same as the score with old drivers and the pre-patched 3dmark03. Also, since NV3x performance in shaders is still poor in every other benchmark, something leads me to believe nvidia is still optimizing...even if there is no visual evidence. Actually, it is still confirmed that nvidia is detecting 3dmark03 in new drivers(44.65).

erm.. just run tests on 5800 Ultra (44.65) with NVAntidetectorscriptD3D (made by unwinder)....

* 3DMark03 (330) score drop 5400 to 3900 marks (IQ doesn't change).

Discussion (http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=13555&pagenumber=3)

[Megatron]

Yep..pretty fishy.

[ChrisW]

Originally posted by StealthHawk
Can nvidia really detect when a screenshot is being taken and re-render it? Maybe I'm missing something.
Absolutely. Remember, nVidia claims a screenshot can not be accurately taken from the frame buffer, so nVidia provided a special algorithm that is used to create a screenshot the way they say is accurate. Remember, these screenshots are not actual screenshots taken from the hardware but are mathematical calcutations of what the screenshot is supposed to look like. Just how hard would it be for nVidia to detect when you are trying to take a screenshot and allow the next frame to use the original shaders? I'm not saying they are, but it is very possible to do just that.