ATi vs nVidia drivers...

[RavenGuard]

Personally, I prefer ATi at this stage in the game. They have some great stuff to offer. A friend of mine disagrees. He's a bit of a nut overall, but I'd be interested in knowing if anyone has ever bothered to do any sort of research on this topic.

His claim is that nVidia drivers tend to improve their cards vastly over time, while ATi cards don't improve much from new software over the same time period.

Any thoughts? I've seen decent improvements with my 4870/4890 with new drivers, but I don't own a current nVidia card to test with. Comments from someone with experience in both series of cards is welcome!

 

[cusideabelincoln]

ATI definitely offers significant improvements in performance over time, especially with their 4800 series cards. Just look at the driver release notes and there are sites who have done some testing. I think even a few members here have done their own testing.

To say whether it's more or less than what Nvidia is something I don't know. That would take a lot of testing in a lot of games to even get a snapshot of the situation. But to say ATI doesn't offer much improvement is just plain ignorance.

Personally, I've never had a problem with either driver set.

 

[Forumpanda]

I wish the driver debate would just die.
No drivers have significant issues, most problems people experience with or attribute to the drivers are unrelated to the drivers, and likely caused by the user.

Every now and again nVidia or ATI releases a driver with a problem for some people, it is inevitable with the millions of different possible system configurations available, but neither side is doing significantly better or worse.

 

[jimhsu]

I own a 8800 GTS 512MB. In my experience, Nvidia drivers only fix problems without affecting performance much, and (with the 190 series) can sometimes cause even more problems. My last ATI card (X1600 mobile) was too underpowered for any performance comparison.

 

[schneiderguy]

I remember some website (I think it was anandtech, but I'm not sure) did a comparison a few years ago on whose drivers improved performance the most. It turned out that nvidia's didn't have significant performance increases over the lifetime of the card, but ATI's did IIRC. This was during the Geforce6 days though, so it's probably not relevant for today.

 

[thilanliyan]

I've also heard of ATI offering more performance improvements over time (ie. X1900 cards fairing better in modern games compared to 7900 cards). Dunno how much truth there is to that as I have not tested it myself.

 

[konakona]

I concur about the driver issue being really overblown. Improvements are good, but not a necessity. People have a random problem or two, sometimes totally minor at that and try to spin it off as a widespread plague. The only thing I noticed was how big Nvidia's driver was getting. Anand's site had an article on it a while ago IRRC, and a big chunk of that was none other than the physx driver.

 

[IlllI]

this topic has never been discussed before

 

[Ben90]

When it comes to optimizations and features they are pretty neck and neck... but when it comes to random misc things like disabling vsync on d3d 7 applications on windows 7...nvidia all the way

(im still looking for a solution to that for my 4890)

 

[schneiderguy]

Originally posted by: thilan29
I've also heard of ATI offering more performance improvements over time (ie. X1900 cards fairing better in modern games compared to 7900 cards). Dunno how much truth

That's probably more due to the architecture of the card than the drivers.

 

[TurtleCrusher]

Both sides get better over time. It's a non issue.

 

[DaveSimmons]

Both have their good days and bad days.

Many of the people claiming driver problems really have unstable overclocks (CPU and/or video card), drivers from ATI/nv installed over the other camp without cleaning, or Windows infected with trojans, rootkits etc. from visiting bad sites and downloading infected games.

 

[cusideabelincoln]

Originally posted by: schneiderguy

Originally posted by: thilan29
I've also heard of ATI offering more performance improvements over time (ie. X1900 cards fairing better in modern games compared to 7900 cards). Dunno how much truth

That's probably more due to the architecture of the card than the drivers.

It took me a few minutes, but I remember that post being made on this forum and I remember AT didn't publish the results. I went through my history and found the link:

http://www.pcgameshardware.com...ations/Reviews/?page=4

http://forums.anandtech.com/me...eadid=2280254#30577861

I would like a little bit more thorough review, however, to make sure some of those "lower end cards" like the 7900 and X1900 actually provide playable framerates. I'm not sure what to take from benchmarks where the 6800 Ultra gets 5 FPS and the X850 gets 10 FPS. Technically the X850 would be twice as fast, but I wonder how the situation would change at lower detail settings and higher framerates.

 

[Grinja]

I believe that the biggest improvements come in the way of better CF support, particulary for the X2 cards.

 

[AzN]

What your friend said has a valid point. Particularly Nvidia improving drivers over time. However Nvidia sometimes come out with drivers that degrades IQ for performance . They've been caught few times in the past and I don't know how the IQ degrades gradually as nvidia optimize drivers.

I've owned ATI cards for generations and same for Nvidia. The difference is there when it comes to image quality and how the frame rates behave in both taxing and non taxing situations.

ATI definitely has the chip now. They always had it but sometimes they brought dumb things like 2900xt.

Which is better? I guess the one that please you.

 

[DaveSimmons]

Both ATI and nvidia have been caught cheating by rendering popular games incorrectly to score better on benchmarks.

Both have their good days and bad days.

 

[RavenGuard]

All right, well it seems that the majority of the comments agree with my "that's BS, they're both more or less equal in terms of improving over time" stance. Any additions are welcome, especially if someone wants to crack out some old hardware and drivers

 

[wanderer27]

Originally posted by: Forumpanda
I wish the driver debate would just die.
No drivers have significant issues, most problems people experience with or attribute to the drivers are unrelated to the drivers, and likely caused by the user.

Every now and again nVidia or ATI releases a driver with a problem for some people, it is inevitable with the millions of different possible system configurations available, but neither side is doing significantly better or worse.

I wish that were true, but it seems nVidia is breaking their drivers these days:

http://forums.ubi.com/eve/foru...301057942/m/6101077677


Why do we have to hack drivers to get a game that's a few years old to work?

I have and have had many nVidia cards over the years, but this is just crap - ATI isn't doing this.
Could it be they're (nVidia) doing this to jack up performance in other areas?

 

[Sable]

ATI have been fucking up their drivers recently. Both 9.8 and 9.9 have spacked up DVB viewer. God knows what they did the AVIVO in it but pretty sure that was the problem. 9.6 mangled the desktop refreshrate so I couldn't use 85hz on my CRT. (fuck off CRT haters)

Don;t get me wrong, 9.7 is excellent, no problems at all. Games all working. DVB Viewer perfect, AVIVO working fine. Refresh rates all fine.

No idea what NV are like these days.

 

[Forumpanda]

As mentioned, its unreasonable to think either ATI or nVidia can test the drivers in every game and on every possible system configuration.
They obviously wont release a driver with know problems (besides the ones mentioned in the driver doc).

There is just a limit to how many tests you can run.
And also, the people with problems obviously are the ones who speak the loudest.

An argument can almost be made that more driver testing would not even benefit consumers (if it meant longer time between each release).
If a fix is created for a problem I have, then I would rather they release it early (with the possibility that something else breaks), if the fix helps me I can upgrade drivers quickly instead of waiting months for them to be 'QA'ed'.

Not that I can ever remember having problems due to a driver upgrade from either side, and I am usually diligent about upgrading with every release.

 

[wanderer27]

Removing a feature, in this case "z-buffer wireframe" , I think is more than just a bug.

To me, this seems to be a deliberate action - it has nothing to do with more testing.