PCPERDX12 GPU and CPU Performance Tested: Ashes of the Singularity Benchmark

[gamervivek]

Gamegpu Test

http://gamegpu.ru/rts-/-strategii/ashes-of-the-singularity-test-gpu.html

It seems 980Ti is leading Fury X in every situation now and sometimes very convincingly, I am glad that the 980Ti is bad at Dx12 myth is busted for good.

I'm not sure what you're being glad about. The big brouhaha about this benchmark was that 390X was like 2fps behind 980Ti in arstechnica's review and they went gaga over it. Secondly, AMD's performance shot up while nvidia's even decreased slightly in dx12. And finally that poster on ocn.net writing a wall of text about how nvidia can't do async compute and it will mean AMD will win big in dx12.

Fury X wasn't scaling well over 390X(here it'd be on par with the nano) or it'd have put some distance between itself and 980Ti and this was noticed even then. So 980Ti besting Fury X wasn't the crux of the issue.

gamegpu results look like heavily gpu bottlenecked with AMD's dx11 performance slightly behind dx12 and not like 20-50% behind.

 

[MagickMan]

Gamegpu Test

http://gamegpu.ru/rts-/-strategii/ashes-of-the-singularity-test-gpu.html

It seems 980Ti is leading Fury X in every situation now and sometimes very convincingly, I am glad that the 980Ti is bad at Dx12 myth is busted for good.

Yep, the anti-NV guys really can't say much anymore. As I thought before, it would only take some additional work with the drivers for the 980TI to get back on top.

All that said, I don't want AMD to die, no competition in GPUs hurts us gamers.

 

[Jaydip]

I'm not sure what you're being glad about. The big brouhaha about this benchmark was that 390X was like 2fps behind 980Ti in arstechnica's review and they went gaga over it. Secondly, AMD's performance shot up while nvidia's even decreased slightly in dx12. And finally that poster on ocn.net writing a wall of text about how nvidia can't do async compute and it will mean AMD will win big in dx12.

Fury X wasn't scaling well over 390X(here it'd be on par with the nano) or it'd have put some distance between itself and 980Ti and this was noticed even then. So 980Ti besting Fury X wasn't the crux of the issue.

gamegpu results look like heavily gpu bottlenecked with AMD's dx11 performance slightly behind dx12 and not like 20-50% behind.

Just look at some earlier posts, it was predicted that NV had already lost the Dx12 battle, the test show nothing can be further from truth.980 Ti is going strong as expected and the supposedly Dx12 dominance of AMD can't be found.

 

[Keysplayr]

I hope nobody is surprised at this..

 

[3DVagabond]

Gamegpu Test

http://gamegpu.ru/rts-/-strategii/ashes-of-the-singularity-test-gpu.html

It seems 980Ti is leading Fury X in every situation now and sometimes very convincingly, I am glad that the 980Ti is bad at Dx12 myth is busted for good.

So when Fury was leading this bench meant nothing but now the 980 ti is the myth is busted for good? All right then. If you say so.

 

[Leadbox]

I hope nobody is surprised at this..

I am, it looks like AMD has improved their DX11 driver for this game to within a frame or 2 of DX12 or gamegpu is doing it wrong

 

[tviceman]

I pointed the performance improvements Nvidia had made when Anandtech tested heterogenous multi-GPU setups. The AMD crowd completely ignored it then, too. Now 980 TI's are going for less than $600 with AAA game and overclock 25%, while the Fury X is going for $649 (current cheapest on newegg), is generally slower out of the box, and only gets a 5-7% overclock.

 

[3DVagabond]

I pointed the performance improvements Nvidia had made when Anandtech tested heterogenous multi-GPU setups. The AMD crowd completely ignored it then, too. Now 980 TI's are going for less than $600 with AAA game and overclock 25%, while the Fury X is going for $649 (current cheapest on newegg), is generally slower out of the box, and only gets a 5-7% overclock.

Nice sales pitch. Completely OT of course, but go ahead with your agenda.

Infraction issued for trolling.
-- stahlhart

 

[guskline]

Here are my results on my 5960x at 4.4Ghz with a single EVGA GTX980TI SC (core of 1102) at Extreme quality and 2560 x 1440p resolution with latest Nvidia drivers:

 

[Keysplayr]

Nice sales pitch. Completely OT of course, but go ahead with your agenda.

Wow. Hey, maybe you should back off on yours a bit.

 

[Silverforce11]

So when Fury was leading this bench meant nothing but now the 980 ti is the myth is busted for good? All right then. If you say so.

They probably disabled Async Compute completely at the request of NV... Notice the lack of discussion from Oxide recently about this topic, gone silent they have. It was quite strange for them to come out all gung-ho, unusual for developers to be so open/frank. PR must have had a chat with the devs.

Then there's Fable, on UE4, with what the devs claimed was ~5% AC usage. Pretty insignificant, yet the Fury X = 980Ti even at 1080p. For an UE4 game, that's unheard of where AMD is usually 25-50% behind.

I say bring on the DX12 era.

 

[tviceman]

Nice sales pitch. Completely OT of course, but go ahead with your agenda.

AGENDAS OMG @#$ How in the world did you get to be so perceptive among video card forums? I guess pointing out quick facts here and there R 4 AGENDAS. So let me include a quick Fury X fact: it comes with WATER and you can sleep better a night knowing your GPU never goes above 65C. Are my points even and balanced now?


AMD is currently not displaying a DX12 advantage. Not in Fable and no longer in Ashes. It'll happen eventually, just keep believing. AGENDAS.

 

[stahlhart]

Stop the fighting, or this thread is getting locked.
-- stahlhart

 

[desprado]

They probably disabled Async Compute completely at the request of NV... Notice the lack of discussion from Oxide recently about this topic, gone silent they have. It was quite strange for them to come out all gung-ho, unusual for developers to be so open/frank. PR must have had a chat with the devs.

Then there's Fable, on UE4, with what the devs claimed was ~5% AC usage. Pretty insignificant, yet the Fury X = 980Ti even at 1080p. For an UE4 game, that's unheard of where AMD is usually 25-50% behind.

I say bring on the DX12 era.

Plz just stop it. No more excuse or justifying.No one disable AC on ATOS and it is time to expect GTX 980 Ti is much better than Fury X.

 

[caswow]

So nobody is noticing amds dx11 perf.? Either AMD did a good Job or their test is crap

 

[guskline]

I've posted my results. Do any other posters on this thread have Ashes of Singularity?

 

[gamervivek]

Just look at some earlier posts, it was predicted that NV had already lost the Dx12 battle, the test show nothing can be further from truth.980 Ti is going strong as expected and the supposedly Dx12 dominance of AMD can't be found.

I told you why this blew up, arstechnica didn't even use a 390X though. It was a 290X.

This is the pic that started it all,

http://cdn.arstechnica.net/wp-conte...-template-final-full-width-3.0011-980x720.png

This then led to speculation why it might be so and Mahigan at ocn.net posted his async theory and it went on from there.

It was later that the oxide dev. in that thread clarified(and reclarified) that they weren't running async compute on nvidia and that it's nvidia who had been in better contact with them than AMD's marketing logo would have you believe.

And,

Saying that Multi-Engine (aka Async Compute) is the root of performance increases on Ashes between DX11 to DX12 on AMD is definitely not true. Most of the performance gains in AMDs case are due to CPU driver head reductions. Async is a modest perf increase relative to that. Weirdly, though there is a marketing deal on Ashes with AMD, they never did ask us to use async compute. Since it was part of D3D12, we just decided to give it a whirl.

http://www.overclock.net/t/1575638/...able-legends-dx12-benchmark/110#post_24475280

 

[Azix]

Users should be submitting their own performance figures for this now. Then we'll know. Those results seem lower than before.

Don't bank on all the progress made being drivers or such. Oxide is clearly willing to make code changes for nvidia's sake. After so long anything could have happened.

 

[Silverforce11]

AMD is currently not displaying a DX12 advantage. Not in Fable and no longer in Ashes. It'll happen eventually, just keep believing. AGENDAS.

You must be one of those people who believe it's absolutely NORMAL for AMD GPUs to perform so well in an Unreal Engine 4 (NV sponsored engine!) game like Fable.

Even 390 > 980 and Fury X = 980Ti at 1080p resolution, which it is normally 15-20% behind.

As for Ashes, the rest of the stack, AMD GCN > Maxwell, except for Fury X vs 980Ti. That's despite Oxide saying AC is disabled.

There's no advantage... yeah right.

 

[desprado]

You must be one of those people who believe it's absolutely NORMAL for AMD GPUs to perform so well in an Unreal Engine 4 (NV sponsored engine!) game like Fable.

Even 390 > 980 and Fury X = 980Ti at 1080p resolution, which it is normally 15-20% behind.

As for Ashes, the rest of the stack, AMD GCN > Maxwell, except for Fury X vs 980Ti. That's despite Oxide saying AC is disabled.

There's no advantage... yeah right.

Can you show any proof of advantage or any screenshot where AMD is having advantage? if not than you are just trolling.

 

[poofyhairguy]

Kepler below the 780/Titan gets crushed in Directx12 mode:

Tahiti gets a nice boost.

 

[Jaydip]

So when Fury was leading this bench meant nothing but now the 980 ti is the myth is busted for good? All right then. If you say so.

This is a GE game so I was not at all surprised when Fury X was leading just like what is happening in SWBF but that was not the main discussion here, it was foretold over and over that NV was doomed during the Dx12 era, so now NV has come out and beaten FuryX in a GE title and in Dx12 mode so I say yeah NV is doing Dx12 all right.

 

[Jaydip]

I told you why this blew up, arstechnica didn't even use a 390X though. It was a 290X.

This is the pic that started it all,

http://cdn.arstechnica.net/wp-conte...-template-final-full-width-3.0011-980x720.png

This then led to speculation why it might be so and Mahigan at ocn.net posted his async theory and it went on from there.

It was later that the oxide dev. in that thread clarified(and reclarified) that they weren't running async compute on nvidia and that it's nvidia who had been in better contact with them than AMD's marketing logo would have you believe.

And,



http://www.overclock.net/t/1575638/...able-legends-dx12-benchmark/110#post_24475280

So the crux of the matter is the issue was blown out of proportion and some people mislead many on purpose, I am glad that it is settled for good.

 

[Erenhardt]

So nobody is noticing amds dx11 perf.? Either AMD did a good Job or their test is crap

Elephant in the room everyone ignores.
AMD dx11 perf improved 50% compared to nv counterparts.

 

[Erenhardt]

This is a GE game so I was not at all surprised when Fury X was leading just like what is happening in SWBF but that was not the main discussion here, it was foretold over and over that NV was doomed during the Dx12 era, so now NV has come out and beaten FuryX in a GE title and in Dx12 mode so I say yeah NV is doing Dx12 all right.

So the crux of the matter is the issue was blown out of proportion and some people mislead many on purpose, I am glad that it is settled for good.

Hold your horses. DX12 is not here yet. We will have multiple discussion on this matter in the future.