Battlefield 1 Benchmarks (Gamegpu & the rest)

[moonbogg]

So what do you think the minimum fps would be with a GTX 1080 if they did do these benchmarks in 64 slot full multiplayer BF1 server ?

I wish they would.

I have no idea. I haven't played the full game yet and haven't really studied the benchmarks. I'll be more motivated once I buy the game. I'm waiting for the official release.

 

[Mercennarius]

Interesting to see the perfect Fury X vs 290X TFLOP scaling.

Is this an end to the "Fiji front end bottleneck" idea and whenever we do not see this scaling it's simply fail drivers? Or are the 290X drivers fail here? I think it's the latter.

I think Hawaii isn't performing here. 290X = 470 isn't quite right. Also 290 is only 4-8% faster than the 380X at 1440p DX11-12. That's called a red flag, folks.

Odd to see Hawaii suffer when it often does the best seeing as its closest architecturally to whats in the consoles.

Look at how Hawaii performed in this comparison:

390X being the 4th fastest GPU in DX12 and out pacing the 980 Ti. Significantly better than the gamegpu numbers. 390X was also 20+% faster than the regular 290.
http://wccftech.com/battlefield-1-directx-12-benchmarks-amd-nvidia/

 

[Spjut]

PCgameshardware's latest update says Nvidia DX12 suffers from a problem similar to the synchronization issue Nvidia had in Doom Vulkan initially, introducing significant latency. They don't say AMD has that issue but is complaining about DX12 performance in general.

Hopefully new drivers fixing this will be available at the official release.

 

[crisium]

Look at how Hawaii performed in this comparison:

390X being the 4th fastest GPU in DX12 and out pacing the 980 Ti. Significantly better than the gamegpu numbers. 390X was also 20+% faster than the regular 290.
http://wccftech.com/battlefield-1-directx-12-benchmarks-amd-nvidia/

That's more in line with what I'd expect to see from AMD. Fury X back to being only 11.4% faster than 390X despite having a 45.4% TFLOP advantage.

480 being ahead of 290 but behind 390X is also what I might expect.

 

[b-mac]

I am pleased to see both companies cards performing well. Shows a job well done by the company that made the game.

 

[Head1985]

Look at how Hawaii performed in this comparison:

390X being the 4th fastest GPU in DX12 and out pacing the 980 Ti. Significantly better than the gamegpu numbers. 390X was also 20+% faster than the regular 290.
http://wccftech.com/battlefield-1-directx-12-benchmarks-amd-nvidia/

390x=GTX1070 now.Funny thing is hawaii started against GTX780.....
It beat GTX780 and then GTX980 and then GTX980TI and now its ready to take down GTX1070.

 

[ThatBuzzkiller]

That's more in line with what I'd expect to see from AMD. Fury X back to being only 11.4% faster than 390X despite having a 45.4% TFLOP advantage.

Weak geometry performance will do that ...

Wasn't exactly a bright idea for AMD to give consoles 2 tri/clock when they have a much lower CU/shader engine ratio configuration like 12/2 or 18/2 compared to Fiji's 64/4 and they dun goofed again with PS4 Pro's 36/4 configuration since it's geometry performance is even more effective with the accelerated primitive discard ...

Hopefully the nightmare for AMD will have ended once devs move to shader model 6, devs focus on project Scorpio so optimizations will be more symmetrical to Fiji, and most importantly they release Vega microarchitecture ...

 

[Azix]

Weak geometry performance will do that ...

Wasn't exactly a bright idea for AMD to give consoles 2 tri/clock when they have a much lower CU/shader engine ratio configuration like 12/2 or 18/2 compared to Fiji's 64/4 and they dun goofed again with PS4 Pro's 36/4 configuration since it's geometry performance is even more effective with the accelerated primitive discard ...

Hopefully the nightmare for AMD will have ended once devs move to shader model 6, devs focus on project Scorpio so optimizations will be more symmetrical to Fiji, and most importantly they release Vega microarchitecture ...

does not compute. fury x is faster than 1070 which I assume has good geometry performance in your view. 390x also has weak geometry performance, probably worse than fury x, so shouldn't that maintain the delta between them then? Or result in no difference at all.

Just seems like diminishing returns in general.

 

[96Firebird]

So who has done multiplayer testing? That is all we should be looking at, the rest are worthless.

 

[Stuka87]

Here is the 4k benchmarks from wcc. It looks almost like the gap between the Fury and the 390X gets smaller.

 

[Flapdrol1337]

So who has done multiplayer testing? That is all we should be looking at, the rest are worthless.

pcgameshardware did a multiplayer test
http://www.pcgameshardware.de/Battl...attlefield-1-Technik-Test-Benchmarks-1210394/
They only tested dx12 in singleplayer though.

The test results for this game are all over the place. 3 posts up someone is praising hawaii, then in other tests hawaii is awful and slower than the RX470

 

[pcslookout]

So who has done multiplayer testing? That is all we should be looking at, the rest are worthless.

I am wondering this too.

We are talking about true multiplayer tests or benchmarks!

 

[ThatBuzzkiller]

does not compute. fury x is faster than 1070 which I assume has good geometry performance in your view. 390x also has weak geometry performance, probably worse than fury x, so shouldn't that maintain the delta between them then? Or result in no difference at all.

Just seems like diminishing returns in general.

Only so much good compute performance will do before AMD hits a wall. Hawaii and Fiji has shown us that in many cases (games), you cannot expect linear scaling in framerates with increase in compute power ...

Nvidia's apporach to GPU design has been far more sensible than AMD's and especially in the last couple of years ...

 

[Bacon1]

Digital Foundry has their initial test up: https://www.youtube.com/watch?v=ZxsIOV2AjMc

DX12 -> 5-30fps faster for 480 in single player

DX11 -> 5-10 fps faster for 1060 in single player

480 keeps above solid 60 (into 100s) @ 1080p. 1060 had some drops under but not sure if that was only in DX12 and if it was solid when in DX11.

 

[nFen]

Digital Foundry has their initial test up: https://www.youtube.com/watch?v=ZxsIOV2AjMc

DX12 -> 5-30fps faster for 480 in single player

DX11 -> 5-10 fps faster for 1060 in single player

480 keeps above solid 60 (into 100s) @ 1080p. 1060 had some drops under but not sure if that was only in DX12 and if it was solid when in DX11.

Nice seeing them use an i5 for testing.

Note that in that video it shows nvidia faster in dx11 @ 2:28. The results are flipped @ 3:57 showing nvidia a couple frames behind. Not too sure what happened there but I suspect that they mislabeled the results in the video.

 

[IllogicalGlory]

Nice seeing them use an i5 for testing.

Note that in that video it shows nvidia faster in dx11 @ 2:28. The results are flipped @ 3:57 showing nvidia a couple frames behind. Not too sure what happened there but I suspect that they mislabeled the results in the video.

The test at 2:28 is DX11 1060 vs DX12 1060. The test at 3:57 is DX11 1060 vs DX12 480.

 

[nFen]

The test at 2:28 is DX11 1060 vs DX12 1060. The test at 3:57 is DX11 1060 vs DX12 480.

I must be seeing things :/

 

[IllogicalGlory]

I must be seeing things :/

I'm sure I've made that mistake once or twice myself.

 

[pcslookout]

Here is a new review!

http://www.guru3d.com/articles-pages/battlefield-1-pc-graphics-benchmark-review,1.html

 

[Head1985]

Here is a new review!

http://www.guru3d.com/articles-pages/battlefield-1-pc-graphics-benchmark-review,1.html

Something is wrong there.
980TI should be faster or at GTX1070level and not 10%faster than GTX970

 

[Mercennarius]

Here is a new review!

http://www.guru3d.com/articles-pages/battlefield-1-pc-graphics-benchmark-review,1.html

Another good showing for the 390X. Ahead of the 980Ti again (in both DX11 and DX12) and on the heals of the regular Fury.

 

[Carfax83]

Another good showing for the 390X. Ahead of the 980Ti again (in both DX11 and DX12) and on the heals of the regular Fury.

A bit premature perhaps? The guy that did that review (Hilbert) said he's going to look into the numbers for the 980 Ti tomorrow, as something is clearly off.

 

[ozmanb]

Something is wrong there.
980TI should be faster or at GTX1070level and not 10%faster than GTX970

You're right, and the reviewer has addressed the concern on the last page of his review. The findings of the 980TI are rather odd. They will re-test the card.

 

[Mercennarius]

A bit premature perhaps? The guy that did that review (Hilbert) said he's going to look into the numbers for the 980 Ti tomorrow, as something is clearly off.

Even if you take the 980Ti out, the 390X still had a great showing.

 

[felang]

Can anybody point me to an article that mentions how the different Video options/levels affect fps/visual quality. I'm running a 290x (1125mhz) and what like to try to optimize vid options in order to maintain close to 60fps @1440p, in other words optimize my config.