Is driver degradation a thing or just made up?

Nexus18 · 25 Jul 2016 at 16:07

I think someone is just a little salty, that is all

spikerules · 25 Jul 2016 at 16:11

Davanius said:
Here's the thing. I didn't read the following post and had no idea that you had posted. I came from the fury vs 980ti thread and thought wow.

There was no need for this.

Well this thread was created after our dispute. I think he wanted more Nvidia fanboy backup, but thankfully we got level heads on both sides

Gregster · 25 Jul 2016 at 16:11

Davanius said:
I hope this thread isn't a reaction to your dispute with spikerules.

Gregster.

It sort of is and there has been a few saying that NVidia are gimping performance and hence me looking and finding this. I see no evidence myself of any purposeful gimping going on. I thought a thread showing some decent testing over many games was a fair call and Kepler would be the one that suffers the most as it is now 2 generations old.

Gregster · 25 Jul 2016 at 16:13

Greebo said:
|Interesting to see the 290x smashing the titan in 10 games there, sometimes by a huge margin.

I suspect overclocked the titan x would pull a few of those round though.

Yer, AMD really did well with the performance improvements and nice going for those who still rock those kinds of GPUs.

Just to add - The Titan was launched Feb 2013 and the 290X launched October 2013, so quite a big gap in launches.

spikerules · 25 Jul 2016 at 16:16

Gregster said:
It sort of is and there has been a few saying that NVidia are gimping performance and hence me looking and finding this. I see no evidence myself of any purposeful gimping going on. I thought a thread showing some decent testing over many games was a fair call and Kepler would be the one that suffers the most as it is now 2 generations old.

Again with this gimping... you seem to not understand the difference between not improving upon older drivers and "gimping"

Gregster said:
Yer, AMD really did well with the performance improvements and nice going for those who still rock those kinds of GPUs.

Just to add - The Titan was launched Feb 2013 and the 290X launched October 2013, so quite a big gap in launches.

Changed your tune there bud... it's okay to be wrong...

Gregster · 25 Jul 2016 at 16:21

spikerules said:
Again with this gimping... you seem to not understand the difference between not improving upon older drivers and "gimping"

Changed your tune there bud... it's okay to be wrong...

I am not wrong though. If you read as many threads and posts as me, you will see the term gimping used often by several different people. This thread isn't all about you lol and it was to show that there isn't any gimping going on. You seem to feel that I am going at you, which I am not and look at the bigger picture and you will see where I am coming from. For the record, when the 290X launched, it was +/- in games as to what was the faster, then AMD improved performance and hence you see the 290X ahead. This is typical of AMD and quite often they release a decent driver that gives the users the performance they paid for.

spikerules · 25 Jul 2016 at 16:25

Gregster said:
I am not wrong though. If you read as many threads and posts as me, you will see the term gimping used often by several different people. This thread isn't all about you lol and it was to show that there isn't any gimping going on. You seem to feel that I am going at you, which I am not and look at the bigger picture and you will see where I am coming from. For the record, when the 290X launched, it was +/- in games as to what was the faster, then AMD improved performance and hence you see the 290X ahead. This is typical of AMD and quite often they release a decent driver that gives the users the performance they paid for.

Yes we see that you have disagreed with me on the other thread and are now basically agreeing with what I said on this one... nice one Gregster. Nice one.

queamin · 25 Jul 2016 at 16:28

I don't think they are gimping but lets be honest i wouldn't put it past Nvidia, but Rroff could answer this wasn't there, about a year ago where people were moaning about Kepler performance on Nvidia forums in certain games and soon after a new driver fixed that or was it a bug?.

Gregster · 25 Jul 2016 at 16:30

spikerules said:
Yes we see that you have disagreed with me on the other thread and are now basically agreeing with what I said on this one... nice one Gregster. Nice one.

No idea how you come to that? Do you even read what I write lol? I can see I am wasting my time though and I won't get dragged in any further, so go with whatever you feel is correct and I will go along with it

Five.Stars · 25 Jul 2016 at 16:31

Greebo said:
|Interesting to see the 290x smashing the titan in 10 games there, sometimes by a huge margin.

I suspect overclocked the titan x would pull a few of those round though.

It is OG Titan. A stock Titan X will obliterate a 290X on every level.

Davanius · 25 Jul 2016 at 16:34

queamin said:
I don't think they are gimping but lets be honest i wouldn't put it past Nvidia, but Rroff could answer this wasn't there, about a year ago where people were moaning about Kepler performance on Nvidia forums in certain games and soon after a new driver fixed that or was it a bug?.

I remember getting rid of my 780 after having excellent performance from launch to pulling my hair out wondering what the hell has happened.

I wouldn't now say it was due to gimping but rather my card being put aside at that time imo.

bru · 25 Jul 2016 at 16:36

This article was posted back in march when it was first released, there was a big thread about it back then as well.

Greebo · 25 Jul 2016 at 16:40

Five.Stars said:
It is OG Titan. A stock Titan X will obliterate a 290X on every level.

Lol I thought that was too good to be true

ANyway here is the best comparison I have ever come across on the Keplar performance as to whether it was gimped or not.

a 780ti tested with 21 driver versions

http://www.bytemedev.com/the-gtx-780-ti-sli-end-of-life-driver-performance-analysis/

As you can see, occasionally there is a "bad" driver which has slightly worse performance than the previous one but overall its a steady climb upwards.

But interesting to see that it only improved 3-4% over 21 driver updates.

Like I said, Nvidia cards tend to be maxxed out driver wise on launch. You will get the odd massive improvement is certain games but thats mainly to the old driver under performing, not a "gain" as such.

By comparison the 290x and hence all AMD cards gained 10-20% during the same time.

arc@css · 25 Jul 2016 at 16:43

A few examples there show great gains, but for the most part (and in general) I don't think it makes all that much difference, really.

Ian Evey · 25 Jul 2016 at 19:46

Seems that some people may not know what gimping means?

pandem0nium · 25 Jul 2016 at 20:01

It's not so much Gimping as planned obsolescence as Matt said the other day.

Amd cards have a lot of potential to unlock over time.

titaniumx3 · 25 Jul 2016 at 20:40

No, it is just AMD users trying to justify the fact that they didn't get full performance from their cards upon release.

Seriously though, I think performance "degradation" is clearly wrong (as shown by the benchmarks posted). However, I wouldn't be at all surprised if nvidia were intentionally not optimizing newer drivers to extract more performance on older hardware. As a business it makes no sense for them to do this, given the lacklustre competition from AMD. Makes far more sense to move their driver support teams to newer hardware. Also, how would anyone even prove that they were doing this?

It's a massive shame that there is so little competition from AMD. Nvidia are probably holding back a fair chunk of performance, both in terms of hardware and software. Remember the GTX280/HD4870 days? Nvidia were actually feeling the burn back then and releasing drivers for the specific aim of optimisng the performance. Nowadays, it's barely a footnote in the driver release notes. Yes sure, they have game ready drivers, but that is not quite the same thing. I won't even go into the pricing, but we all know how awesome that was.

Telecaster · 25 Jul 2016 at 20:48

If you want the fastest card on release day then nvidia have this sewn up.

It's not hard for AMD to improve old cards with new drivers when they're starting from a lower performance to begin with. If you're the sort on a budget who keeps a card for years on end then go AMD, one day you'll get the card you paid for on release.

Griffildur · 25 Jul 2016 at 21:12

oh great, yes let's debate this one more time because the previous 100k times were not enough.

Bored now, you folks keep arguing about this, while I go enjoy a game on my performance degradable 1080 which in 10 years will only be capable to run Solitaire ( if I am lucky! ) and overpriced GSync screen.

See ya!

Panos · 26 Jul 2016 at 00:06

Lets not forget that since 16.3, both with 16.5 and 16.7 all AMD cards got a hefty improvement also. (can see it on my Nano).
More so for the RX480 getting 6% across the board!!!!! between 16.6 and 16.7.2

The unified GCN idea AMD has paid off. Same applies also to the Async compute, or DX12 support from before DX12 was invented!!!!!
290/290X came out before DX12 was even announced.
While I remember here huge arguments that the GTX970/980 are fully compatible DX12 cards because the box sticker stated so, and AMD is all crap when they announced full DX12 support with drivers update only......