Gears of War 4 Benchmark

[96Firebird]

Looking at PCWorld's results, my card is on par with 1060/480? Not bad...

 

[ConsoleLover]

GameGPU Results Out!

Edit: They didn't use NVIDIA release drivers for the game.

http://gamegpu.com/action-/-fps-/-tps/gears-of-war-4-test-gpu

Why don't they test the RX 480? In every review they make they include the 1060 6gb, but not the RX 480 8gb, why is that???

Anyways good scores for Nvidia definitely, this is an Unreal Engine title, which usually has 20% faster results on Nvidia cards, so to see DX12 implementation equalizing things a bit for AMD is good.

Unfortunately this is going to be it going forward, massively skewed engines for Nvidia or AMD, where one company artificially has much better performance.

It will be interesting to see the upcoming DX12 releases most of whom are sponsored by AMD and see how AMD favored engines perform.

 

[Carfax83]

Since when async only increases performance by 2-3% ? oh wait UE4 nvidia's poster child which is fully optimized for nvidia just like oxide was amd's. but as long as nvidia wins nothing matters. all those reasons like "optimized for one vendor, sponsored, bla bla" will be irrelevant now. 970 matching furyx says alot about which hardware this game likes and is optimized for.

This is going to be the most importantest dx12 game ever, just note it down.

Breathe, before you have a meltdown. PCgameshardware.de did some testing on the performance impact of tiled resource and asynchronous compute, and you can bet that AMD increased performance substantially from asynchronous compute:

Fury X got about 15% extra for minimums and 10% increase for average. GTX 1080 about 1% for minimum and 3% for average. It doesn't matter though, because Pascal is WAY ahead of Fury X, and so is the GTX 980 Ti for that matter, which doesn't have AC...

 

[Madpacket]

Perhaps I'll buy this game and do my own testing, UE games generally disappoint though. I'm getting a little tired of seeing sites like Gamers Nexus spew out questionable results and coming off as impartial. One of the only respectful public facing benchmark guy got bored and is on a hiatus (AdoredTV), can't really blame him.

 

[Sweepr]

4GB Fury X with the same min FPS as an 8GB RX 470 @ PCGH, well behind 6GB GTX 980 Ti:

 

[Phynaz]

Since when async only increases performance by 2-3% ? oh wait UE4 nvidia's poster child which is fully optimized for nvidia just like oxide was amd's. but as long as nvidia wins nothing matters. all those reasons like "optimized for one vendor, sponsored, bla bla" will be irrelevant now. 970 matching furyx says alot about which hardware this game likes and is optimized for.

This is going to be the most importantest dx12 game ever, just note it down.

You realize this is a Microsoft game running on Microsoft (AMD) hardware? How you can construe that as optimized for Nvidia I don't know.

 

[Final8ty]

Breathe, before you have a meltdown. PCgameshardware.de did some testing on the performance impact of tiled resource and asynchronous compute, and you can bet that AMD increased performance substantially from asynchronous compute:

Fury X got about 15% extra for minimums and 10% increase for average. GTX 1080 about 1% for minimum and 3% for average. It doesn't matter though, because Pascal is WAY ahead of Fury X, and so is the GTX 980 Ti for that matter, which doesn't have AC...

How does getting an 15% extra for minimums and 10% increase for average does not matter to gamers who have those cards ?

 

[Madpacket]

4GB Fury X with the same min FPS as an 8GB RX 470 @ PCGH, well behind 6GB GTX 980 Ti:

Cool, another area where 4GB doesn't matter unless you like playing at less than peasant speeds.

 

[Madpacket]

You realize this is a Microsoft game running on Microsoft (AMD) hardware? How you can construe that as optimized for Nvidia I don't know.

I agree but it's somewhat interesting this is a game that comes free with Nvidia cards at the moment. Hitman also came as a free game for AMD cards and was clearly optimized for GCN. It's not much of a stretch for Nvidia ensuring a promotional game is very well optimized for their hardware regardless of the console origins.

 

[kraatus77]

You realize this is a Microsoft game running on Microsoft (AMD) hardware? How you can construe that as optimized for Nvidia I don't know.

I realize it's an nvidia sponsored game made on UE4 which is fully optimized for nvidia hardware on pc. i also realize we are looking at pc benchmarks and not so called Microsoft 's hardware (console).

something which was famous on this forums during aots, but now irrelevant because for obvious reasons.

 

[Arachnotronic]

I agree but it's somewhat interesting this is a game that comes free with Nvidia cards at the moment. Hitman also came as a free game for AMD cards and was clearly optimized for GCN. It's not much of a stretch for Nvidia ensuring a promotional game is very well optimized for their hardware regardless of the console origins.

The point here is that those individuals who ran around saying "NVIDIA sux at DX12, AMD is da bomb at DX12" didn't take into account the fact that some of these early DX12/Vulkan titles were simply optimized for AMD's architecture. We are seeing with GoW4 and other DX12 titles that the shoe can very well be on the other foot.

I am not trying to argue that AMD is "better at DX12" or NVIDIA is "better at DX12." What I am trying to argue is that different titles with different engines/architectures may favor one architecture over another. There will be some that favor AMD, others that favor NVIDIA.

 

[Phynaz]

I realize it's an nvidia sponsored game made on UE4 which is fully optimized for nvidia hardware on pc. i also realize we are looking at pc benchmarks and not so called Microsoft 's hardware (console).

something which was famous on this forums during aots, but now irrelevant because for obvious reasons.

So you think the PC version is differently optimized than the console version? Do you have a shred of evidence of that?

 

[Headfoot]

So you think the PC version is differently optimized than the console version? Do you have a shred of evidence of that?

uhh... it must be by default. It doesn't use the same APIs (e.g. DX12 on Xbox is not the same as DX12 on PC). Hardware architecture is fundamentally different with no unified memory pool on PC. No eDRAM/eSRAM on PC, so that has to be optimized differently. PC has slow system memory and fast video memory which must be accounted for in optimization. Saying that the PC version is not optimized any differently than a console version is flat out absurd. Unless you're talking about their optimization goals being different between platforms vs. whether they actually had to code it differently?

 

[PhonakV30]

Review from computerbase is Up.

https://www.computerbase.de/2016-10/gears-of-war-4-benchmark

First frametime of GTX1060 vs RX 480

 

[master_shake_]

So another game that bust both the DX12 and Kepler driver myth.

290x still faster than a 780ti.

when 780ti was released it was way better than a 290x.

also, wow look at that multi-gpu scaling.

or lack there of.

 

[AtenRa]

Review from computerbase is Up.

https://www.computerbase.de/2016-10/gears-of-war-4-benchmark

Its funny to see NVIDIA having CPU overhead in DX-12, AMD cards are way better with slower CPUs this time.

 

[crisium]

But is it on Steam? Unless it's dirt cheap I'm not touching that Windows Store even if it gets 144fps maxed on my card.

 

[dogen1]

Yeap, this is why GTX 970 with only 3.5GB + 0.5 is faster

What other explanation could there be for the 390X having higher 1%s and 0.1%s and a lower average?

 

[Carfax83]

Review from computerbase is Up.

https://www.computerbase.de/2016-10/gears-of-war-4-benchmark

Wow, computerbase.de is the only website which shows the FuryX ahead of the 980 Ti. And not only that, the FuryX is ahead of the 1080 when paired with a FX-8370

Not sure what's going on with their review, but this is totally different from every other reviewer at the moment..

 

[Carfax83]

How does getting an 15% extra for minimums and 10% increase for average does not matter to gamers who have those cards ?

Where did I say it didn't matter for the people that own those cards? I was obviously talking from the context of competition with NVidia..

 

[PhonakV30]

Wow, computerbase.de is the only website which shows the FuryX ahead of the 980 Ti. And not only that, the FuryX is ahead of the 1080 when paired with a FX-8370

Not sure what's going on with their review, but this is totally different from every other reviewer at the moment..

Well , Is there any review with FX-8370 ? atm only Computerbase.Also, GameGpu shows that only at 4K Fury X is faster than GTX 980 TI.mind that both are running at Reference clock.

Don't forget that GameNexus shows that Fury X Is equal to GTX 970!

 

[kraatus77]

So you think the PC version is differently optimized than the console version? Do you have a shred of evidence of that?

Face palm.

 

[AtenRa]

Wow, computerbase.de is the only website which shows the FuryX ahead of the 980 Ti. And not only that, the FuryX is ahead of the 1080 when paired with a FX-8370

Not sure what's going on with their review, but this is totally different from every other reviewer at the moment..

If im not mistaken they are the only one or one of the few who benchmark an actual game play and not an in-Game Benchmark.

https://translate.google.com/translate?sl=de&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=https://www.computerbase.de/2016-10/gears-of-war-4-benchmark/&edit-text=

The test sequence lasts 25 seconds and is one of the more challenging passages in the game. It shows a walk with many explosions and various partying effects.

 

[Stormflux]

Not sure what's going on with their review, but this is totally different from every other reviewer at the moment..

You were so quick to defend Gamersnexus on the first page. ALL outliers should be further examined. Not just the ones that suit agendas.

http://www.gamersnexus.net/site-news/2629-gears-4-testing

We have discovered a few issues with the Gears of War 4 testing that require a revisit to the game. We are working diligently to perform those tests now, and have temporarily unpublished the original content while we work to learn more about the title.

Our apologies for the inconvenience as we work through some new tests with the game. These are important to the results, and we believe them to be critical enough to put a pause on our original content delivery.

 

[Carfax83]

If im not mistaken they are the only one or one of the few who benchmark an actual game play and not an in-Game Benchmark.

PCgameshardware.de doesn't use the in game benchmark either, and their results are completely different.