Gameworks, Mantle and a pot calling a kettle black

Martini1991 · 12 Jun 2014 at 14:22

LtMatt said:
I don't really know how to answer your question. All i can do is explain why AMD are able to catch up when X8 AA is applied due to fact GCN handles AA better than Kepler. I still don't understand how a 770 is able to best a 290X though when AA is not applied. Can you explain that?

You haven't really explained diddly.
People know that Kepler takes a bigger hit than current AMD GCN when using MSAA.

However, the drop is never as large as we see in Batman AO.

My question did "explain/address" (Not that it's definitive, nor am I claiming it to be fact) but is it not possible than Nvidia are simply running far higher than they should be, while AMD is running as they should be, not slower than they should be? It's the only way to possible explain why Nvidia's losing so much using MSAA, we're talking like a 40 FPS lead turning into a 2 FPS lead.

LtMatt · 12 Jun 2014 at 14:27

Martini1991 said:
You haven't really explained diddly.
People know that Kepler takes a bigger hit than current AMD GCN when using MSAA.

However, the drop is never as large as we see in Batman AO.

My question did "explain/address" (Not that it's definitive, nor am I claiming it to be fact) but is it not possible than Nvidia are simply running far higher than they should be, while AMD is running as they should be, not slower than they should be? It's the only way to possible explain why Nvidia's losing so much using MSAA, we're talking like a 40 FPS lead turning into a 2 FPS lead.

That can only come down to their drivers then. Nvidia do most of their optimization through their drivers according to the recent GameWorks article. Perhaps their AA optimization needs work, or perhaps they need to work more with the dev. AA is something AMD were able to optimize at driver level in this game. According to that article AMD work closely with the devs to implement their optimizations into the engine and their drivers, a 50-50 collaboration while Nvidia is more like 80-20. The dev (warner brothers) refused any of amds optimizations, because of GameWorks. Either they weren't allowed to help AMD or they didn't have the licence to look and or alter the source code. No way to prove either way which one it was as they refused to comment on it.

Gregster · 12 Jun 2014 at 14:27

LtMatt said:
Yes and i said i changed my mind somewhat on what GameWorks intention was. Not sure what else you want from me really.

As I've already explained at x8 AA AMD cards are able to overpower any advantage that is present when only FXAA only is used, for whatever reason. Under no circumstance is it normal for a 770 to be beating a 290X though, i think we can all agree that.

Nope, I have massive respect for what the 290/X can do but that is just wrong. If the gap is that big with FXAA, it would be that big with 8XMSAA or as near as damn it, as shown with all the other bench threads.

LtMatt · 12 Jun 2014 at 14:30

Gregster said:
Nope, I have massive respect for what the 290/X can do but that is just wrong. If the gap is that big with FXAA, it would be that big with 8XMSAA or as near as damn it, as shown with all the other bench threads.

It happens across every card though Greg, its not just unique to the 290X. All the AMD cards get a big boost (in performance comparative to the Nvidia cards) when AA is applied and a massive performance hit when FXAA is used.

Admiral Huddy · 12 Jun 2014 at 14:30

Matt, If you don't want people to have an opinion then best not to start such an inflammatory thread or post which is likely to come back this. Likewise, I've seen you do very much the same thing elsewhere so the thread title is pretty accurate.. IMHO

TheRealDeal · 12 Jun 2014 at 14:31

Gregster said:
Nope, I have massive respect for what the 290/X can do but that is just wrong. If the gap is that big with FXAA, it would be that big with 8XMSAA or as near as damn it, as shown with all the other bench threads.

No it wouldn't be as the 290x is obviously taking a big hit using fxaa. You would need to see what both cards were getting with no fxaa to see the real difference between them at 8xmsaa. For all we know the 290x is getting more frames with no fxaa v the Titan with no fxaa.

LtMatt · 12 Jun 2014 at 14:32

Admiral Huddy said:
Matt, you have reported various posts in this thread for off topic and trolling.., If you don't want people to have an opinion then best not to start such an inflammatory thread such as this. Likewise, I've seen you do very much the same thing elsewhere so the thread title is pretty accurate.. IMHO

This is not my thread.

Gregster · 12 Jun 2014 at 14:34

LtMatt said:
It happens across every card though Greg, its not just unique to the 290X. All the AMD cards get a big boost (in performance comparative to the Nvidia cards) when AA is applied and a massive performance hit when FXAA is used.

I am not disputing that AMD cards handle MSAA better than nVidia but I am disputing the jump. No way would that kind of jump happen with MSAA and from my own experience with nVidia, the move from FXAA to MSAA is about right (not with scientific testing of course).

Gregster · 12 Jun 2014 at 14:35

LtMatt said:
This is not my thread.

I am guessing he means the other GameWorks threads you started.

LtMatt · 12 Jun 2014 at 14:35

Gregster said:
I am not disputing that AMD cards handle MSAA better than nVidia but I am disputing the jump. No way would that kind of jump happen with MSAA and from my own experience with nVidia, the move from FXAA to MSAA is about right (not with scientific testing of course).

I think the problem is actually the performance hit AMD cards take with FXAA rather than the performance hit Nvidia take with x8 AA.

Admiral Huddy · 12 Jun 2014 at 14:37

LtMatt said:
This is not my thread.

.. granted this time but he others.. but you know what i mean regardless.

LtMatt · 12 Jun 2014 at 14:42

Admiral Huddy said:
.. granted this time but he others.. but you know what i mean regardless.

I don't mind a good debate, but im not fond of the baiting or name calling.

triss · 12 Jun 2014 at 14:43

Martini1991 said:
You haven't really explained diddly.
People know that Kepler takes a bigger hit than current AMD GCN when using MSAA.

However, the drop is never as large as we see in Batman AO.

My question did "explain/address" (Not that it's definitive, nor am I claiming it to be fact) but is it not possible than Nvidia are simply running far higher than they should be, while AMD is running as they should be, not slower than they should be? It's the only way to possible explain why Nvidia's losing so much using MSAA, we're talking like a 40 FPS lead turning into a 2 FPS lead.

Same thing many have touched on Gameworks is simply optimizing Nvidias cards higher then where they would normaly be . Not saying this is exactly what happens but its is what gameworks really should do

Gregster · 12 Jun 2014 at 14:44

LtMatt said:
I think the problem is actually the performance hit AMD cards take with FXAA rather than the performance hit Nvidia take with x8 AA.

So back to the original points from long ago, could it just be AMD have not optimised fully for FXAA in this game? At max settings and with AMD handling 8XMSAA better than nVidia, I see no other reason why the massive drop when MSAA is changed out for FXAA in honesty. Like I say, with my own testing in plenty of games (and at 4K), Batman AO pretty much is in line with other AAA titles I have tested with FXAA, so seems spot on to me for nVidia cards.

llamadal · 12 Jun 2014 at 14:44

unless it's you doing the baiting?

Martini1991 · 12 Jun 2014 at 14:46

Simple solution, someone benches Batman on AMD using no AA, and then using FXAA.
If there's no massive performance gain from not using FXAA (Assuming it can be turned off) then it's not FXAA doing the gimping?

TheRealDeal · 12 Jun 2014 at 14:47

Martini1991 said:
Simple solution, someone benches Batman on AMD using no AA, and then using FXAA.
If there's no massive performance gain from not using FXAA (Assuming it can be turned off) then it's not FXAA doing the gimping?

+1 i have been looking around a bit trying to find this very scenario but can't.

Gregster · 12 Jun 2014 at 14:48

LtMatt said:
Vs you, Nvidia fanboy and GameWorks defender. Who to believe... game devs for me.

LtMatt said:
I don't mind a good debate, but im not fond of the baiting or name calling.

That was pretty harsh in my opinion.

LtMatt · 12 Jun 2014 at 14:49

Gregster said:
So back to the original points from long ago, could it just be AMD have not optimised fully for FXAA in this game? At max settings and with AMD handling 8XMSAA better than nVidia, I see no other reason why the massive drop when MSAA is changed out for FXAA in honesty. Like I say, with my own testing in plenty of games (and at 4K), Batman AO pretty much is in line with other AAA titles I have tested with FXAA, so seems spot on to me for nVidia cards.

As FXAA is a post process effect i don't think it requires any optimization at driver level as it is applied after the work is done. I think you're barking up the wrong tree there.

llamadal said:
unless it's you doing the baiting?

I love a good debate, but i try not cross the line by insulting people or persistent baiting as has been seen against me for the past few pages. There's a definite difference i think. I've stepped over the line previously and taken deserved punishment for it.

Gregster said:
That was pretty harsh in my opinion.

I get called a fanboy every day here.

Fanboy is being a fan of something, not necessarily a bad thing i don't think. So far ive been told i talk bullcrap, im utterly clueless and have been baited continuously for the past couple of pages. Slight difference i say.

Gregster · 12 Jun 2014 at 14:50

Martini1991 said:
Simple solution, someone benches Batman on AMD using no AA, and then using FXAA.
If there's no massive performance gain from not using FXAA (Assuming it can be turned off) then it's not FXAA doing the gimping?

Good call and it can be turned off. The problem is finding someone with AMD who owns Batman AO