• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Nvidia theory. Is Nvidia purposely doing this

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Mikesamo

Member
Joined
Jun 10, 2014
Location
san jose CA
What are your thoughts on this.
My friend tested a titan x maxwell against a fury x back during its release and the titan won in everything. Fast forward to today and on metro 2033 they are almost the same on par in performance. Do you believe Nvidia purposely gimps their older generation cards to boost their sales on newer ones . Or is it just a matter of driver support priority
 
I don't believe they purposefully gimp older cards. I do believe that AMD doesn't release cards with their full potential due, at least partly, to under optimized drivers for the card(s) performance. Once they get the drivers updated enough times there tends to be some performance boosts. Although, to be fair you're benchmarking a 6 year old game.
 
There's a good chance that Nvidia just don't care, just like Radeon, when they were ATI, back in 2004, 2005 and maybe 2006.

Reminds me of 10 years ago, if not a little later, that at least some things here and there seemed to have been gimped with Catalyst versions later than 3.2 with Radeon 9000 Pro. But, some tests were better with the later Catalyst drivers for that period...
It seemed that only something here and there would have a regression...

On top of that, there were unconfirmed reports of the latest Catalyst drivers of that period locking out video card OC'ing! And at that time, never heard of Nvidia doing that kind of thing!

Back then, Nvidia drivers were KISS! ATI's latest of the time, compared to Nvidia's, were bloated!

Sadly, it looks like Nvidia is heading down the same road that ATI was notorious for... (At least the benchmark regressions with later driver versions...)

I was wondering if my GeForce GTX 660 has any regressions with the latest driver...

I doubt a GTX 1060 would this year, lol.
 
Last edited:
you might be right. i put my old 460 back in the pc just for lols and i feel like its weaker than i thought. my standards changed so much since going to 1440p 144 hz from 1080p 60hz
 
AMD is improving drivers for all, new and older generations while Nvidia is improving performance in new drivers only for current generation. When GTX900 was released then it was ~20% faster than GTX700 series. In next couple of weeks difference went up to 50% in some titles ( comparing similar priced shelves ). Since GTX1000 was introduced, GTX900 got no new performance improvements.
Other thing is that AMD is usually getting more with driver updates. Look at RX480 vs GTX1060. Released at about the same time and after first comparisons RX480 was 5-20% slower in many games. Now is up to 15% faster in some of them. Still there are more games that run faster on Nvidia but GTX1060 isn't so much faster than after its premiere while RX480 looks much better right now.
 
Fury X drivers have improved heaps and the card is basically on par with the 980ti.

"Overall, the GTX 980 Ti is still significantly faster than the Fury X in the majority of our games" - notice the term used "significantly".

AMD seems to gain much more speed overall then nVidia as drivers progress as Woomack said. I switched from 362.00 to 375.95 and didn't really see any gains except in a few select new games that came out in the meantime.
 
Last edited:
They have improved, no doubt. But your own link, Saul, says its still falling well short of the 980Ti...

To add/finish to what Kenrou said/saying:
Overall, the GTX 980 Ti is still significantly faster than the Fury X in the majority of our games although the Fury X has been able to gain a little ground – 7 additional benches out of the 75 we originally tested are now in the Fury X’ favor – although it is still bested by the GTX 980 Ti overall.
 
"Overall, the GTX 980 Ti is still significantly faster than the Fury X in the majority of our games" - notice the term used "significantly".

AMD seems to gain much more speed overall then nVidia as drivers progress as Woomack said. I switched from 362.00 to 375.95 and didn't really see any gains except in a few select new games that came out in the meantime.

edit looking further into it the fury X is pretty far behind although improvements have been made. (lol I browsed way too quick to come to my conclusion that the fury X was basically as good as the 980ti - apologies)

The fury X is probably not the best example of an AMD card gaining performance over time I'd say the 7970 and 290X are much more interesting and the 480 has gained a lot in 4 months (5%-8%).
 
Last edited:
If someone had far too much time on their hands, they could try downloading historic drivers from a GPU's launch to current, install each one and run a load of benchmarks along the way. Has it been done already? Otherwise, any volunteers?
 
If someone had far too much time on their hands, they could try downloading historic drivers from a GPU's launch to current, install each one and run a load of benchmarks along the way. Has it been done already? Otherwise, any volunteers?

There's a YouTube user called Pato 9 that does nVidia driver reviews/benchmarks since 361.91, he's Brazilian though so PM if you need translation.
 
If someone had far too much time on their hands, they could try downloading historic drivers from a GPU's launch to current, install each one and run a load of benchmarks along the way. Has it been done already? Otherwise, any volunteers?

I have a 560ti and a 460 laying around i suppose ill volunteer . can someone lend me their non gsync monitor
:D
 
Not to go off topic but my curiosity kicked in and wants to ask, why is it that some drivers effect Folding? I notice the warning of the past and as of recent actually, for 10xx users complaining that their PPD decreased and that goes for older gens too. IIRC, I understand that the work units you get become more "advanced" overtime so it takes longer to complete for a older GPU. Does that make any sense or am I blowing up smoke? lol

I brought this up to maybe see if there was any shred of evidence of the OP's theory. For the record, I don't completely believe they do but a small part of me does. lol
 
For the same reason it effects FPS. A driver is, essentially, a how to. How do I 'crunch' 'this'....
 
Exactly. Driver version affects benchmarks. They always have. Why would you think it wouldn't affect crunching also?
 
AMD is improving drivers for all, new and older generations while Nvidia is improving performance in new drivers only for current generation. When GTX900 was released then it was ~20% faster than GTX700 series. In next couple of weeks difference went up to 50% in some titles ( comparing similar priced shelves ). Since GTX1000 was introduced, GTX900 got no new performance improvements.
Other thing is that AMD is usually getting more with driver updates. Look at RX480 vs GTX1060. Released at about the same time and after first comparisons RX480 was 5-20% slower in many games. Now is up to 15% faster in some of them. Still there are more games that run faster on Nvidia but GTX1060 isn't so much faster than after its premiere while RX480 looks much better right now.

So would you recommend getting a 480 over a 1060? I have a an i5-6600 coupled with R9 280X, gets me all ultra @ 1080p with only occasional dips below 30 fps, but @1440p (VCR) I'd probably be down to 25fps average. Figured I don't desperately need an upgrade but it wouldn't hurt either. After a couple of issues with AMD — poor support, lying, setting retailers up, deleting feedback on social media etc. — I promised them my next card would be nVidia. Now I'm having second thoughts. This said, right now 1060 tends to be at least 10% cheaper in Poland at least.

No, I do not think it is done purposely. I think it is a bi-product of natural progression.

And perhaps some added confusion resulting from changing corporate attitudes and policies.

Drivers operation is all Chinese to me. Can someone briefly explain how they work

Long story short, nobody really knows, and different individual cards work better with different individual versions of all-card drivers. Newest drivers aren't always best.

Insiting on having one and the same set of drivers for all cards is bad IMHO, as improving performance for one card gimps another, so it's always a give-and-take, which is stupid and unfair. Not receiving improvements or continued support is one thing, but having your performance sacrificed (even as a by-result) so that a more recent customer could get a boost should be a no-go zone. As should be software blocks. Drivers may have EULAs prohibiting you from doing certain things, but actual cards are physical objects sold unconditionally, and copyright/patent/IP protection shouldn't be abused to put a card buyer at the manufacturer's mercy. Cards aren't just licensed, they're owned by their buyers.
 
Last edited:
Back