Does nvidia use driver updates to kill performance on ageing GPUs? Linus says no.

Thoughts on his (actually Jake's) research?

 

Thoughts? For the people who actually believed this was a thing, I've got a bridge to sell you.

 

I have to agree. This wasn't a very thorough test. He should have tested drivers pre new architecture launch. That's when we see gimping take place.

 

There is no gimping. It's always the people who never provide proof that claim gimping, while those of us that benchmark games regularly after driver updates see that driver performance trends upward over time or stays the same.

 

I was wondering about this myself. Testing a singe post-launch driver doesn't seem comprehensive enough to draw much of a concrete conclusion from.

I've seen respected members of this forum have different views on this. I also am not one who regularly tests games after driver updates, though.

@Papusan @Phoenix - as our regular driver testers/egg cookers, do you have any thoughts on whether nvidia uses driver updates to gimp old GPUs?

 

It doesn't matter what your view is. The evidence is there. Driver performance goes up or stays the same over time. I'm talking in games, I could care less about synthetics. The biggest gains from driver optimization come in the first few months to a year after the release of a new GPU uarch before tapering off. AMD usually gains more from driver optimization than Nvidia (so-called FineWine™ Technology) because their GPUs are released in a less optimized state than Nvidia's, and because of their presence in console silicon which affects the development of game engines and emergent rendering techniques that favor their architecture.

This is like the second video LTT has done on this nGimpia conspiracy theory. The entire premise is so ridiculous I have a hard time taking it, and people who actually believe it, seriously. I have a feeling that I've been using PC for longer than some of these people have been alive, so I remember when actual controversy existed in the GPU space, rather than this FUD. Like drivers being coded to skip the rendering of some elements in a certain popular benchmark to inflate scores for one side, or a certain popular game locking out a rendering path on one side due to that side's terrible performance in that API, or notable performance and image quality differences in basic things like anti-aliasing and anisotropic filtering between the two.

 

Well this has been the latest trend for the past 2 years. When approaching the release date of a new GPU Drivers start getting worse and worse and have problems making people feel sick of them and eventually end up upgrading. That's how I was forced to upgrade from a Dual 780Ms on my Alienware 18 to a Dual 980M GTX on a Clevo laptop back then

Sometimes they would just fry your GPU, much easier than pissing customers off for a long time. Always best to stick to the drivers from the OEMs that they test unless a new driver has some game optimizations which you are playing

 

I think he meant dual 780m.

Thinking back I'm reminded of how Nvidia loaded up tesselation on Maxwell release in gameworks to make AMD cards run like ass, even to the detriment of nuking Kepler GPU performance. That's acceptable collateral damage for them.
Then recently with the Titan, the newer magic driver that conveniently unlocks magic compute powers ahead of AMD 64.
Nah Nvidia are a sly bunch.

 

Those are separate issues from "Nvidia releases newer drivers to gimp performance of games on older GPUs", which has been proven as false so many times. In other words, in the same games, older drivers DO NOT perform better than newer drivers.

 

just consumers being whiney babies cos their gpu can't run the latest games maxed out

 

Nvidia gimping old arch GPU performance via Gameworks is move appropriate then. I've not seen any compelling proof dispelling overloaded tessellation via gameworks around Maxwell release as mentioned above. If you've got some I'd be interested to see it

 

That's the game developer's fault, not Nvidia's.

I'd like to see your proof of newer drivers harming performance on the same GPU(s) in the same games compared to older drivers.

 

Good point. Half the people crying gimp are the Pentium and 750 Ti users trying to run Ghost Recon: Wildlands at max settings.

 

It's both their fault. Developers save time and money and Nvidia get to put whatever they please into closed off code.

I never claimed I had proof of this nor stated that is emphatically what's happening.
I'm saying Nvidia pull shenanigans at convenient times.
I'm open to evidence of either side. Hence why I asked you to share it since you claimed said shenanigans have been proven false multiple times.

 

Then please don't obfuscate the topic of the OP to use this as a platform to bash Nvidia for other unrelated things. The topic is whether or not Nvidia gimps performance of older GPUs via driver updates. And the answer is no, performance goes up or stays the same. There is one piece of evidence in the OP's video, here is another video by Linus from 2 years ago that shows the same thing on a Fermi GTX 480. I have yet to see anyone in the nGimpia camp post similar sort of evidence to back up their hogwash, just a bunch of hearsay and anecdotes, hence my question to you.

 

You'll never get evidence to the contrary, because there is none. It's just a tin foiled boogeyman.

 

I had my 870M for three years I never noticed any degradation in performance.

Games got more hungry little by little. Of course it's dependent on the game really. Some devs pay more attention to their games performing better on weaker hardware than others.

 

Nvidia is just optimizing their latest gen of GPUs for new games. Older gen's usually fall behind and that's why people do believe they crippled the performance of older cards which is obviously not true. AMD on the other hand does more rely on hardware (raw GPU power) than squeezing the last bit of perf. out of their driver. That's why their GPU's constantly gain perf. over time, as developers become more familiar with the architecture and drivers mature. A good example for that is the GTX 970 vs R9 390...