-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Gameworks, Mantle and a pot calling a kettle black
- Thread starter Gregster
- Start date
Soldato
- Joined
- 27 Feb 2012
- Posts
- 6,586
In defence of driver optimisation, i'm rocking the old 13.12 drivers. So if Martini is running 14.6, if he runs the same @1100/1350 that will show how much has been improved.
I mean, it's either drivers or CPU as we're neck and neck on 8xMSAA score front.
Edit: Or Matt, or any 290 owner....I mean, most of you are against GW for gimping your performance. Help try to prove them wrong/right. Unless you have a hidden agenda
I mean, it's either drivers or CPU as we're neck and neck on 8xMSAA score front.
Edit: Or Matt, or any 290 owner....I mean, most of you are against GW for gimping your performance. Help try to prove them wrong/right. Unless you have a hidden agenda
Last edited:
Soldato
- Joined
- 6 Feb 2010
- Posts
- 14,583
IF GW libraries is same as libraries in general that people understand to be that is. But as it stands, there's no conclusion on where it went wrong, and neither AMD's claim of GW making it difficult for them to optimise for the game or Nvidia's claim of GW doesn't hurt AMD's performance has solid evidence to back it up. There's one fact though...the 256-bit 32ROPs 7850 got its rear handed to it by a far lower spec 128-bit 16ROPs GTX650Ti, so something went seriously wrong somewhere...
GW is an API like I've specified before, you can see the documentation here: http://docs.nvidia.com/gameworks/index.html
As for the claim that it might be more "tightly integrated" into games, why then does the Unreal 4 engine come without GW but has very simple "plug and play" support for GW if it needs to be tightly integrated?
Well as proven by Tone and I's testing there is the same drop off in performance moving from no AA to FXAA. So I think AMD just didn't have the driver optimised properly for its card at the time of the initial reviews.
They probably weren't allowed access till late on but who knows?
That's a whole can of worms that i don't want to get into.
At the end of the day AMD have not been able to look at the source code so there is not much they can do. Devs say that is vital and often mandatory. Nvidia admit that you need to do that if there is a problem. Middleware (unoptimised) is buggy and problematic. If the dev is not willing to help them or modify things, there is absolutely nothing they can do. You see where this is leading? I think this is why you have these problems for AMD in GW titles.
With all that said and done, I now no longer think that Nvidia created GameWorks with a view to gimping AMD performance. I've changed my opinions a bit from previously thanks to article written by Joel but mainly due to words coming from the developers. They have no reason to lie to us, or tell half truths unlike AMD/Nvidia.
Edit: Or Matt, or any 290 owner....I mean, most of you are against GW for gimping your performance. Help try to prove them wrong/right. Unless you have a hidden agenda
As I've said previously above.
Just need a 770 user to post the results now and see if it's faster or not than a 290.
Exactly the same as Tomb Raider. nVidia didn't get access till later on but in nVidia's defence, they didn't hang about and got them done quickly.
Soldato
- Joined
- 29 Aug 2010
- Posts
- 8,638
- Location
- Cornwall
I don't think anyone actually agrees that Nvidia's lock out of PhysX if there's an AMD card is a good thing. So using it as a rebuttal can get moot, because they're in agreement anyway
But mmj's comment there is far too blanket lol.
I'm an AMD user with spare Nvidia cards sitting around. I'm not saying I think it's a good thing, but I don't think it's a bad thing either.
Regarding the Batman: AO tests, I did notice that FXAA seemed to hit Minimum FPS quite hard, on Greg's Titan it didn't.
With regards to the 7850 getting owned by a 650Ti (and I believe you've also said a 560Ti, same card rebrand?), which drivers were the 7850 tests done on? One review I found seemed to show considerable gain with the 13.11 beta6 drivers over the 13.11 beta1 drivers.
Missing the point - with no AA to FXAA we both drop off by 3-4%. FXAA doesn't harm AMD any more than nVidia. The raw performance data is irrelevant because you've then got other variables coming in to play. If AMDs performance isn't up to scratch then that says to me that their driver team did not optimise properly.
Whatever the reason for the deficit in performance, it's not because AMD have not done the best they can to optimize performance. To think they would not even attempt to do that is daft and i think you know better than that Rusty. It's not as if Batman is not a popular game so why bother.
So we know it's not FXAA causing it, but theres no way we can prove what is causing it. It could be GameWorks, it could just be a game optimization issue. I don't think we'll ever know the real reason. None of us can say with certainty what the reason is or isn't.
Soldato
- Joined
- 27 Feb 2012
- Posts
- 6,586
You've lost your mojo Matt, you've been so confident because Twitter and Google tell you it's GW causing AMD's lackluster performance. It's been somewhat debunked, and now you want a 770 to perform the test?
All a 770 will show you is exactly what we already know, FXAA isn't hampering performance, it's just AMD cards perform that poorly. I'm on 13.12 driver, play along and test it with your 14.6 drivers.
You've lost your mojo Matt, you've been so confident because Twitter and Google tell you it's GW causing AMD's lackluster performance. It's been somewhat debunked, and now you want a 770 to perform the test?
All a 770 will show you is exactly what we already know, FXAA isn't hampering performance, it's just AMD cards perform that poorly. I'm on 13.12 driver, play along and test it with your 14.6 drivers.
The only reason people think that, is because these issues only crop up in GameWorks titles. That was the whole issue everyone had with it though. How can a mid range gpu beat a top range gpu? Has there every been a game where a 7970 has beaten a 780TI at 1080P? How can a budget nvidia gpu (650TI) beat out a mid range 7870 AMD gpu? I think it only seems the sensible option to test a 770 and see how much faster it is. So far we've discovered that fxaa is not the culprit. We all assumed it was that because of all the benchmark results that showed the difference between x8 AA and FXAA. No one knew for sure. All we know now is FXAA is not the cause for the performance deficit.
It happens. Sometimes games perform badly in comparison on one set of hardware. Nothing more; nothing less.
It's up to the driver teams to resolve it. Perhaps AMD need some code entered into the game to get to nVidia level performance and WB refused it? Perhaps AMD aren't that bothered about optimising for it. Just speculating...
Man, you won't here a thing bad said against AMD and this is what is stopping you from seeing the big picture and making your vision blinkered. You are still banging the "Could be GameWorks causing it" when clearly everything shown to you disproves that and shows it "could" in no way harm AMD performance.
It happens. Sometimes games perform badly in comparison on one set of hardware. Nothing more; nothing less.
Maybe so, but i think given the difference in results between AA on and FXAA is enough for people to wonder what is happening.
Can you cite any examples where a 7970 beats a 780TI at 1080P? I've had a good look but can't find any.
Soldato
- Joined
- 27 Feb 2012
- Posts
- 6,586
Could be because Nvidia's driver teams are on the ball and offer performance as it should, stop looking for someone else to blame other than AMD. Could be, looking at the past this may well be the case. Just speculating though, don't have any twitter feeds to prove it, obviously.
It's nothing to do with a bad word said against AMD. Their optimization process at game and driver level explicitly relies on working with the developer to improve performance. To think they would not even bother is daft. The interview with game devs taught us that about the different methods both sides use to optimize performance. I'm keeping an open mind. Can't say for certainty GameWorks is causing it, or not causing it. Nothing we have seen so far proves it is or isn't. As Marine rightly said, none of us know how deeply it's integrated into the engine.
It isn't relevant.
As I said: sometimes a game runs badly on a set of hardware. If AMDs customers focused their annoyance at AMD instead of misdirected fire anywhere but AMD then perhaps things would improve.
In absence of any other reason in the comparative lack of performance you have to ask questions of the driver team. Just be realistic about it instead of looking for reasons of anything BUT AMD. If you aren't of that opinion then I have to say you're giving that impression.
Could be. Could be any number of reasons. I can't say for sure what the reason is and neither can you.