nVidia release new all singing all dancing dets.

[Sabastian]

After reading a lot of forums, these drivers appear to do the following:
-increase 3dmark scores in most high end systems by up to 10% most notably in the nature test
-sometimes increase fps with high end cards in high quality settings with a few games
-sometimes with some games increase AA peformance of high end cards
-sometimes reduce performance in a few games on various systems
-in most games on most systems performance change stays within 1%
-generally reduce image quality in most games
-disable bi/trilinear filtiring at default settings
-in some cases reduce image quality so much that games become unplayable
-generally reduce AA quality in most games
-change the control panel and add options
-sometimes fix and sometimes create refresh rate issues

Summary: drivers do increase performance on high end systems/cards in a few situations, most noticably in 3dmark. Performance increase in games in general is quite rare and sometimes the performance is even reduced. In general, performance in most games and most situations is about the same. Image quality is lowered quite a bit in a lot of games, same goes for AA quality.

Forums i've checked: arstechnica, rag3d, overcockers.com, hardocp, beyond3d and tweakers.net.

Well I hate to repost this but I feel it is most relevent in this thread.

"Nvidia Playing the 3DMark Game?

http://www.3dchipset.com/news.shtml#newsitem1030639337,5718,

Nvidia Playing the 3DMark Game? - Thursday, August 29 | Solomon

Reports are coming in that the supposed, "Up to a 25% increase in performance" is only showing in 3DMark2001 benchmark. While playing games, Quake 3 is appearing darker then usual. More and more reports will be coming in regarding the new Detonator 40 driver. From the looks of it, it's not what people where really interested in as you can't play 3DMark2001.

More reports to come...

UPDATE: http://www.3dchipset.com/news.shtml#newsitem1030653283,62055,

Nvidia Detonator 40 Results - Thursday, August 29 | Solomon

I received word from VR Zone that they have tested a couple of games as well as the ever popular 3DMark2001. A minor improvement in games as typical with a new revision release, but the increase in 3DMark2001 is some what questionable.

I'm going to jump on a limb here and say what I have to say because most web sites usually will not talk about Nvidia. I believe these drivers are very controversial. Funny how the only program that gets a huge gain is a native benchmark used across many sites as the definite benchmark. Is this just a mere coincidence? I highly doubt it, but alot of people won't say shit about it. Now if ATi did this sort of thing, sites across the net would just be analyzing and picking the driver set apart so they can point the finger. Just think about this. Compare the 9700 pro reviews at launch to say Nvidia video card launches, then you can basically come to the conclusion that alot of sites are padded by Nvidia. Say what you will, but it's quite obvious that only a few forum users are complaining about the huge increase in the popular 3DMark2001 benchmark program and not in their games that they are playing in. Oh well.. You either bendover and take it up the brown eye and enjoy it or be taken to the back room and suffer the hammer to the fingers! I wonder if ATi got used to the hammer? :-\

UPDATED: http://www.3dchipset.com/news.shtml#newsitem1030658608,48961,

More Detonator 40 Results - Thursday, August 29 | Solomon

I swear Nvidia should just rename this release as, "3DMark2001 Detonators". More results are pooring in and it's basically the same as the other one from VR-Zone. These are only for 3DMark2001. Seems Nvidia just wants to boost up their 3DMark2001 score to help out it's cause against the ATi Radeon 9700 pro. Will it be enough to persuade the viewers? PC Extreme has their results up in German. So for those folks who can't read German bust out the translator if you want to read the words from the benchmarker/reviewer. Or just stare at the charts and results as those are universal."

 

[Galilee]

I wish people could focus on drivers and graphics instead of PR. Well, whatever rocks your boat. (not sure it's written what way).

If you wan't to make sure the IQ is the same as earlier drivers make sure to enable Texture Sharpening (under FSAA), and use 1X Anisotropic filtering.

 

[Snap]

I wish people could focus on drivers and graphics instead of PR. Well, whatever rocks your boat.

Tell that to NVIDIA, because this release is nothing more then PR.

 

[Galilee]

Who the fu** cares? (beside fanATIcs)
Do you know that there are atleast ten threads about these NVIDIA drivers at Rage3d? Some of them are over ten pages long. I think 3dgpu have one thread or something (a few pages).

Obviously these drivers are still "beta".

 

[McElvis]

I think they are listed as Beta as they have not been Microsoft certified

 

[tEd]

Actually the default setting is "application preference".
3D Mark doesn't have AF settings, but in applications that does AF can be enabled.

Once you touched the AF slider you can no longer reset this default behaviour other than uninstalling the driver. (That's what I did - and I'm running 30.82 again...)

thats' not true. you just need to push "restore defaults" (d3d driver panel) to set aniso 0 again to applications preference

 

[Bigus Dickus]

So is anyone going to actually post screenshots of Nature and advanced pixel shader tests with 30.xx and 40.xx drivers for us to compare?

Shouldn't that settle the IQ (cheating) question?

 

[Neeyik]

Shouldn't that settle the IQ (cheating) question?

There's no "cheating" to settle. Go back a few pages and you'll see that I've posted test results of Nature with the 30.30s and 40.41s - not surprisingly the latter are considerably higher. I then posted scores of all the tests in 3DMark with the AF setting at 0x and 1x (and yes, it looks rough in 0x)...result? No difference - which should come as no surprise if people actually remembered their hardware stats. Each pipeline TMU is designed, on a GF4 at least, to fetch a bilinearly filtered texel in single clock cycle in the first place (and so memory bandwidth hit has been taken into consideration), so what is there to be gained by using nearest-point?

 

[Reverend]

Wow, this thread is still growing!

1) Install drivers, do not tweak the control panel and run 3D Mark. Although the control panel setting is at 0x, 3D mark runs with 1X (bilinear) setting.

You sure 1x = bilinear and not as per application setting?

BTW, I've noticed some random corruptions with F1 2002 and 3DMark. The effect is akin to overclocking the memory too much, resulting in odd white flashes once in a while and is totally random. Never happened with older drivers before.

 

[Neeyik]

You sure 1x = bilinear and not as per application setting?

As far as 3DMark is concerned, the AF settings of 1x and "application setting" ARE the same as it only uses bilinear filtering.

 

[KimB]

BTW, I've noticed some random corruptions with F1 2002 and 3DMark. The effect is akin to overclocking the memory too much, resulting in odd white flashes once in a while and is totally random. Never happened with older drivers before.

I noticed that too, Rev. It appears that every once in a while, the whole screen flashes (So far, only in 3DMark2k1 on my comp...don't have F1 2002). Obviously a driver bug.

 

[Galilee]

But what is this Texture Sharpening option? Looks like 2X aniso, or a slightly sharper LOD. Was it previously off? Or has it been disabled now?

Reverend: How about you making a in-depth article about these drivers? With performance, anisotrpic tests and FSAA tests. Some people report better IQ in FSAA with these drivers.

A fancy little article from you could settle all these discussions

 

[Reverend]

Chalnoth, I'm not too sure it's a driver bug. The way it (i.e. the white flashes) looks to me doesn't look like a driver bug. It could be, um, "pushing things too far" so to say.

Galilee, as interesting as such an article may be to many folks, it doesn't sound like an exciting proposition to me... so I'll pass. I'm just looking forward to playing with a R9700 and, later (as well as hopefully ) a NV30.

 

[KimB]

Chalnoth, I'm not too sure it's a driver bug. The way it (i.e. the white flashes) looks to me doesn't look like a driver bug. It could be, um, "pushing things too far" so to say.

We'll see, but it seems too regular to be akin to memory overclocking artifacts. I've never seen an overclocking artifact that affected the whole screen for a single frame. Mostly I've gotten corruptions that affect pieces of the screen at once.

 

[Reverend]

Oh and as for the Texture Sharpening, I believe this has been available for quite some time (i.e. many driver sets ago) but had been "hidden". I can't confirm this but I honestly believe that it is simply actually forcing aniso (and not messing with the LOD bias). Would be interesting to bench using Texture Sharpening (without AA... and you can enable texture sharpening without enabling any AA if anyone isn't aware of this due to texture sharpening appearing in the AA control panel) and 2xAniso and comparing the results (I'd think the results would be similar).

 

[Reverend]

Chalnoth, I'm not too sure it's a driver bug. The way it (i.e. the white flashes) looks to me doesn't look like a driver bug. It could be, um, "pushing things too far" so to say.

We'll see, but it seems too regular to be akin to memory overclocking artifacts. I've never seen an overclocking artifact that affected the whole screen for a single frame. Mostly I've gotten corruptions that affect pieces of the screen at once.

Hmm... then we have different observations. The corruptions do not affect the "whole screen for a single frame". In my case, it affects only portions of a frame, random in terms of when it happens as well as where it happens on-screen-within-the-frame. Very much like memory overclocked too much. In 3DMark, I noticed it in only the Nature scene.

 

[flf]

I'm getting both whole-screen corruptions (Dark Age of Camelot character selection screen) as well as partial-screen screwups (Daoc char screen and in Heroes of Might & Magic 3), as well as flickering mountains in the distance in Daoc itself.

Definitely driver bugs. I haven't had these types of problems in any of the previous dets.

 

[McElvis]

BTW, I've noticed some random corruptions with F1 2002 and 3DMark. The effect is akin to overclocking the memory too much, resulting in odd white flashes once in a while and is totally random. Never happened with older drivers before.

I noticed that too, Rev. It appears that every once in a while, the whole screen flashes (So far, only in 3DMark2k1 on my comp...don't have F1 2002). Obviously a driver bug.

Interesting...I've had this problem before on my GF2. I think it was ~21.23 version.

I've also seen this on my friends GF3 as well.

 

[jb]

So anybody else have any educated guess the reason for the increase in the pixel shaders test? Or is it probably some optimizations that were done?

 

[Reverend]

So anybody else have any educated guess the reason for the increase in the pixel shaders test? Or is it probably some optimizations that were done?

Er, has it been confirmed that the improvements are for the pixel shaders and not the vertex shaders? Insofar as 3DMark (Nature) and Aquanox are concerned, I don't think we can separate the two (vertex and pix shaders).

In any case, your second question probably answered your first, ben .