AMD silently nerfing the performance of older GCN cards

[Syntaxvgm]

6 hours ago, Hunter259 said:

Again. No proof of that. Only if by "gimped" you mean slowed driver support then yes. But thats too be expected when you completely change an architecture and how the GPU renders.

I specifically remember my 970 outperforming the 780ti on w3 launch, and driver updates making it worse for a while. Meanwhile I could make it playable on a 6870 at medium 1080p 30fps. 

https://www.reddit.com/r/pcmasterrace/comments/36ysva/witcher_3_on_a_really_old_amd_card_youre_all/

 

I didn't realize it until after that thread, so I tried a gtx 580 and couldn't make the game playable at launch. That's fucking backwards, and while i think it's somewhat been fixed, that is not the only time NVIDIA has done that. Best part, I noticed downgrading the drivers to pre-witcher 3 launch seemed to make the performance much better on the 580, and I heard it was true on the 780ti as well. And, at that time, the 980ti was launched along with the w3 drivers which made your driver crash often on the desktop (games were fine) and this wasn't fixed for months. A 970 owner like myself at the time could roll back, but 980ti owners had to suffer. 

[Demonking]

1 hour ago, LAwLz said:

 

I am 100% sure that this thread would have looked very differently if it said "Nvidia disables <insert feature here> on GTX 600 cards".

Boy you ain't never lied. 

[leadeater]

I have to wonder if there is a legitimate reason to disable this feature on GCN 1.0, which if there is should have been made VERY clear when doing so. It may have been found that while AMD themselves initially thought it was 100% compliant many developers might have been reporting issues with cards based on this architecture and having to do specific tweaks/disables to solve them, an unusable buggy feature might as well be disabled as it saves everyone the hassle.

 

Also could be that continuing support on GCN 1.0 is hindering future development and utilization of async-compute due to hardware limitations only found in this architecture with no clean way to distinguish between these and newer cards, driver reports the feature is supported so all cards are equal right? (Yes that was a simplistic generalization)

 

There is no point in complaining about a feature getting disabled if it was never able to be utilized in the first place, either being no performance gain or it never really worked anyway. Think of it like the old "Turbo Button" on Pentium computers, it NEVER did anything and anyway it was originally designed to slow systems down not speed them up lol.

[Prysin]

4 hours ago, samcool55 said:

AMD changed the names from GCN 1.X to X somewhere during the announcement stuff of polaris.

 

So instead of polais having GCN 1.4 it has GCN 4.

 

Actually The Whole GCN 1.x Was The Tech Press's Doing.  AMD Always Talked Of "next Gen GCN",  But Didn't Actually Name Them.  Was Some Article About It A While Before Polaris Launch

[jimakos234]

1 hour ago, TheRandomness said:

Yet the latest drivers will apparently allow for wattman to be used with the 200/300/Fury series?

Only the R9 285 was specified out of the 280 series, that is a tonga gpu, and the 380 that is basically the 285. No tahiti cards were mentioned at the new driver announcement, so not many of the new goodies will be available for these older cards.

[Kinda Bottlenecked]

Ayyye why AMD?

 

Too bad I didn't believe nvidia nerfed their older gen cards. And now i can't talk smack about amd.

[Sakkura]

2 hours ago, LAwLz said:

Even if we assume that was true, it still does not address all the other hypocrisies I brought up.

Hypocrisies or strawmen? There are far more Nvidia defenders than AMD fans on this forum.

[Sakkura]

2 hours ago, Kwee said:

You misunderstand or the website make you believe that but that not the truth.

 

You're the one misunderstanding. What's he's saying is the reason AMD disabled it (in the driver, affecting all games) is apparently that it ran into issues in that particular game.

 

If that is in fact the explanation, it's an understandable but still not great move. Fix a critical issue in one game while perhaps losing a bit of performance in a lot of other games...

[Kwee]

5 minutes ago, Sakkura said:

You're the one misunderstanding. What's he's saying is the reason AMD disabled it (in the driver, affecting all games) is apparently that it ran into issues in that particular game.

 

If that is in fact the explanation, it's an understandable but still not great move. Fix a critical issue in one game while perhaps losing a bit of performance in a lot of other games...

I'm the one who write the original post, what he saying doesn't exist. Oxide never communicate on that and the desactivation of Async Compute appear recently in newest drivers. 

[TheRandomness]

1 hour ago, jimakos234 said:

Only the R9 285 was specified out of the 280 series, that is a tonga gpu, and the 380 that is basically the 285. No tahiti cards were mentioned at the new driver announcement, so not many of the new goodies will be available for these older cards.

I beg to differ. 

[jimakos234]

7 minutes ago, TheRandomness said:

I beg to differ. 

where do you see tahiti in this photo? keep in mind 285!=280/280X/7970/7950

290/390 hawaii

380/285 tonga

r9/r7 260/360 bonaire

you only confirmed what i was saying, that tahiti is not included

 

[ivan134]

13 minutes ago, TheRandomness said:

I beg to differ. 

Might wanna look again homie.

[Trik'Stari]

AMD...... Don't start with this bullshit. I was going to give you a chance come the 490 and or the Fury equivalent........

[Equlix]

#amd #likeafinewine

seems like just yesterday amd was the budget hero for increasing 480 performance... oh wait.

[S w a t s o n]

Jesus christ, I can't trust any hardware vendor these days. Why the flying fuck would you do this AMD, at least Nvidia's conspiracy for nerfing cards wasn't so obvious. Someone needs to start a 3rd vendor and just not do anti-consumer shit like this.

 

Edit: If they had to do this for driver or bug reasons I guess it's passable but honestly they should be able to issue seperate driver forks (or internally distinguish them) if need be. I have a feeling this isn't the case...

[Carclis]

Really disappointing to see if this is true. I don't mind older cards being sidelined to make it easier for new cards to get the frequent updates, but some GCN 1.0 cards are just over a year old...

[Bouzoo]

2 hours ago, Carclis said:

Really disappointing to see if this is true. I don't mind older cards being sidelined to make it easier for new cards to get the frequent updates, but some GCN 1.0 cards are just over a year old...

You have mixed there something. The last 1.0 cards were 280/280X. 

[Space Reptile]

13 hours ago, Doobeedoo said:

Hm was it a certain reason something. I have R9 290 and didn't see any difference.

290 is gcn2 and thus unaffected ,

as I sayd earlier gcn1 cards don't even update to 16.9.2 only to 16.7.3 and thus are not really that affected irl , only if you run the latest driver it is not going to update natively you will "nerf" the card , in like 2 games

[Space Reptile]

9 hours ago, AresKrieger said:

Think of it this way, how would people react if Nvidia blocked something like tesslation features on the maxwell cards, thus lowering performance 20%

PhysX requires Cuda3 or higher , wich means 600 or 700 series cards or higher , async requires GCN2 or higher , wich means R9 290 or R9 300 series and up

[MarvinKMooney]

OMG a company trying to get people to buy newer stuff that's outrageous, not surprised nvidea does it so why not not like other companies do the same thing.

[TheRandomness]

6 hours ago, ivan134 said:

Might wanna look again homie.

 

6 hours ago, jimakos234 said:

where do you see tahiti in this photo? keep in mind 285!=280/280X/7970/7950

290/390 hawaii

380/285 tonga

r9/r7 260/360 bonaire

you only confirmed what i was saying, that tahiti is not included

Ok, I'll give you that, it was about 2am when I posted that  But I still disproved your last point, saying 'not many of these goodies will be availible for older cards'. It's pretty much everything that wasn't a part of the original 7000 series launch (which was 2012~, the bonaire chip was launched in 2013).

[Carclis]

1 hour ago, Bouzoo said:

You have mixed there something. The last 1.0 cards were 280/280X. 

Well the 370/X along with the 350 and below are GCN 1.0. I'm sure there wouldn't be too many of those out there but it's still pretty unfortunate.

[Prysin]

9 hours ago, Kwee said:

I'm the one who write the original post, what he saying doesn't exist. Oxide never communicate on that and the desactivation of Async Compute appear recently in newest drivers. 

https://mobile.twitter.com/BitsAndChipsEng/status/806617175037513728

 

you sir have some explaining to do. And also some drivers to uninstall with DDU

[Prysin]

6 hours ago, Swatson said:

Jesus christ, I can't trust any hardware vendor these days. Why the flying fuck would you do this AMD, at least Nvidia's conspiracy for nerfing cards wasn't so obvious. Someone needs to start a 3rd vendor and just not do anti-consumer shit like this.

 

Edit: If they had to do this for driver or bug reasons I guess it's passable but honestly they should be able to issue seperate driver forks (or internally distinguish them) if need be. I have a feeling this isn't the case...

trusting any corporation is the biggest fallacy right after believing the earth is flat and that there is a massive secret patriarchy actively trying to oppress and rape western first world women

[AlTech]

10 hours ago, leadeater said:

I have to wonder if there is a legitimate reason to disable this feature on GCN 1.0,

I think that GCN 1.0 cards have bad asynchronous compute support since IIRC, GCN 1.1 was touted as having vastly superior Asynchronous Compute.

10 hours ago, leadeater said:

Also could be that continuing support on GCN 1.0 is hindering future development and utilization of async-compute due to hardware limitations only found in this architecture

Kind of. AFAIK Developers who optimize for AMD specifically in their games (like Hitman) need to write slightly different code for each architectural overhaul which specifically references the architecture and it's features.

10 hours ago, leadeater said:

with no clean way to distinguish between these and newer cards

AMD's Dev tools easily differentiates them and their drivers enable specific features depending on what GCN architecture version they are using.

 

For example, GCN 1.3/3 (whatever you want to call it) has improved tessellation. Meanwhile GCN 1.0 may have worse tessellation performance.

1 hour ago, Carclis said:

Well the 370/X along with the 350 and below are GCN 1.0. I'm sure there wouldn't be too many of those out there but it's still pretty unfortunate.

Everybody should be thankful that they discontinued non GCN cards with the 300 series.