Battlefield 1 Benchmarks (Gamegpu & the rest)

[Grooveriding]

First benchmark from GameGPU

http://gamegpu.com/action-/-fps-/-tps/battlefield-1-test-gpu

DX11

DX12

 

[Annisman*]

Thx Groovey poopypie wonder why there's no Titan X in there ? Also, for the performance drop in DX12 are we getting any extra visual fidelity ?

 

[raghu78]

OP please add 4k charts . the Rx 480 in DX12 is on par with GTX 980 in DX11 at both 1080p and 1440p. Rx 480 DX12 is 17% faster than GTX 1060 DX11 at 1080p and 10% faster at 1440p. RX 470 DX12 is faster than GTX 970 DX11. Nvidia cards lose perf a bit in DX12 while AMD cards gain a bit.

 

[Bacon1]

Odd there is no CFX support in DX11 or any MGPU for DX12. Wonder how they tested so much hardware

CPU Scaling

DX11:

DX12:

 

[Bacon1]

4K very high quality settings:

 

[IllogicalGlory]

Will future DICE games (Battlefield 1) also benefit/need 8GB+ VRAM VGAs to run at max settings?

Nope.

I don't like Intel CPUs losing performance in DX12 though.

Maybe they aren't losing performance. Maybe the test is GPU limited.

 

[Erenhardt]

Big RED flag for 290/x performance. DX12 helps, but still... hoped to see better numbers from hawaii.

 

[Bacon1]

Nope.

Yeah I was shocked by how bad the perf was for Mirrors Edge compared to BF since they use the same engine. Still need to complete Mirrors Edge at some point hope they give it a DX12 update like BF1

 

[Bacon1]

Big RED flag for 290/x performance. DX12 helps, but still... hoped to see better numbers from hawaii.

43min 55avg @ 1440p max settings isn't good? :O

Game seems to run at or better than BF4 while looking even better

 

[Mercennarius]

Big RED flag for 290/x performance. DX12 helps, but still... hoped to see better numbers from hawaii.

Do they use reference 290X cards or non reference? The difference between a reference 4GB 290X vs an overclocked non reference 8GB 390X can be 20% or more in some games...

 

[RussianSensation]

Looks like NV hasn't optimized DX12 drivers or didn't bother as their DX11 performance is good.

R9 380X is 32% faster than GTX960 4GB.
RX 480 beats both the 1060 and 980.

RX 480 is 22% faster than GTX1060 at 1440p DX12.
R9 290 beats 780Ti by 10%.
Nano beats 980Ti and is 33% faster than 980. Where are all the Nano haters who said "we've had that level of performance when 980 came out?"

Perfect scaling from R9 290X to Fury X - 53%. (2816 SP x 1000mhz vs. 4096 SP x 1050mhz)

Under DX11, i7 2600K > i5 6600. HT FTW! i7 2600K has 28% higher minimums than 2500K.

DX12 gives a nice boost to AMD CPUs. A lot of people only focus on how DX12 impacts GPU performance and ignore that DX12 was also meant to improve CPU multi-threaded performance and lower the CPU draw call bottleneck.

Under DX12, i7 2600K > i5 6600. i7 2600K now has 38% higher minimums than 2500K. For nearly 6 years we've heard how HT is useless for games and time after time its benefits in real world benchmarks keep getting ignored. No wonder i7 2600K users are reluctant to upgrade to Skylake i5s.

Battlefield 1 and Star Wars Battlefront are the epitome of GPU optimization --- the games run well on both AMD and NV hardware, scale well with faster NV/AMD GPUs; all while the games are maybe the best looking PC titles in all of 2015-2016. This is night and day compared to the optimization (or lack thereof) we've seen from recent Xbox One to PC ports.

 

[Erenhardt]

Nope.

I don't like Intel CPUs losing performance in DX12 though.

Maybe they aren't losing performance. Maybe the test is GPU limited.

Nope, that is nv GPU giving up under dx12.

43min 55avg @ 1440p max settings isn't good? :O

Game seems to run at or better than BF4 while looking even better

It runs very good, but I was talking about performance relative to other cards. It may be because of this:

Perfect scaling from R9 290X to Fury X - 53%.

 

[pcslookout]

Thanks.

Going to go buy a Nvidia GTX 1070 unless I should wait?

 

[Vaporizer]

RX470 outperforming 1060. . Seems like 1060 is already on the way down the ladder.

 

[daxzy]

Thanks.

Going to go buy a Nvidia GTX 1070 unless I should wait?

Unless you need it for release, might as well wait for some potential holiday sales. Nothing new is coming out until next year.

 

[RussianSensation]

Thanks.

Going to go buy a Nvidia GTX 1070 unless I should wait?

1070 is a good card no doubt but for Battlefield 1, since it's so well optimized, you can buy a manufacturer refurbished Sapphire Fury for $225 (after $25 off with MasterPass checkout). Nothing will beat this deal for Battlefield 1.

1070 is finally coming down in price. 4 months after launch it's now possible to buy an AIB 1070 for $380. If BF1 is your main game, I don't see how it's worth it though over the $225 Fury.

 

[AtenRa]

Hmm Fury NANO 175W TDP 28nm = GTX 1070 150W TDP 16nm at 1440p DX-12

Insane perf/watt for 28nm product.

First Gen GCN are not doing all that well vs Kepler in DX-12, but its like the second coming for Tonga, nice perf in DX-12.

 

[RussianSensation]

RX470 outperforming 1060. . Seems like 1060 is already on the way down the ladder.

Best BF1 GPU Under $300? | AMD RX480 vs GTX 1060

1060 loses 25% of its performance (96 fps DX11 vs. 77 fps DX12) with Joker's testing when moving from DX11 to DX12. The performance is still excellent on both the 1060 and RX 480 at DX11 or DX12 compared to all of the recently ported Xbox One games.

First Gen GCN are not doing all that well vs Kepler in DX-12, but its like the second coming for Tonga, nice perf in DX-12.

Or is it Tonga that's doing exceptionally well?

1080p DX11
$299 R9 280 = 54 fps (GCN 1.0 wins)
$380-450 770 = 50 fps

1080p DX12
R9 280 = 49 fps (GCN 1.0 wins)
770 = 47 fps

Now look at Tonga

1080p DX11
R9 380X = 64 fps (Tonga wins)
GTX780 = 63 fps
GTX960 = 55 fps

1080p DX12
R9 380X = 66 fps (Tonga wins)
GTX780 = 58 fps
GTX960 = 50 fps

GCN1.0 R9 280X is faster than GTX770 Kepler despite costing less for most of the time during that generation. Seems to me like Tonga is actually punching way above its weight as its approaching R9 290.

 

[Bacon1]

1070 is a good card no doubt but for Battlefield 1, since it's so well optimized, you can buy a manufacturer refurbished Sapphire Fury for $225 (after $25 off with MasterPass checkout). Nothing will beat this deal for Battlefield 1.

1070 is finally coming down in price. 4 months after launch it's now possible to buy an AIB 1070 for $380. If BF1 is your main game, I don't see how it's worth it though over the $225 Fury.

Yep, Fury Air performance will be right between Nano and X which places it basically right at 1070 level, for way way cheaper.

 

[pcslookout]

1070 is a good card no doubt but for Battlefield 1, since it's so well optimized, you can buy a manufacturer refurbished Sapphire Fury for $225 (after $25 off with MasterPass checkout). Nothing will beat this deal for Battlefield 1.

1070 is finally coming down in price. 4 months after launch it's now possible to buy an AIB 1070 for $380. If BF1 is your main game, I don't see how it's worth it though over the $225 Fury.

Which uses less power/electricity GTX 1070 or Sapphire Fury ?

 

[RussianSensation]

Which uses less power/electricity GTX 1070 or Sapphire Fury ?

25W of power difference on average between the Fury and AIB 1070.

At peak, 32W difference

One card costs $380, the other $225. In 18 months you can sell the Fury for $140 and get a much faster GPU for $400 when Volta releases in 2018. You are betting that overpaying upfront to future-proof with the 1070 over the Fury is worth $155 spent today as opposed to $155 put aside towards a 2018 Volta level card after reselling the Fury. That's a losing bet because in just 2 years $550 980's level of performance can be purchased in a $190 GTX1060. In 1.5 years, $650 980, $650 Fury X and $550 Fury have dropped to $350, $325 and $225-250. You think GTX1070/1080 won't suffer the same fate in 1.5-2 years from now? History repeats.

 

[Snarf Snarf]

Which uses less power/electricity GTX 1070 or Sapphire Fury ?

The 1070 will use less, but I don't think your power bills will make up $155 over the lifetime of your card to compensate for the huge price difference.

Edit. RS beat me to it with the clutch graphs. It really is negligible amount of power consumption difference.

 

[pcslookout]

Thanks RussianSensation very interesting!

Is there any advantage a GTX 1070 has then?

I found a eVGA Video Card GTX 1070 ACX 3.0 8GB DDR5 256Bit PCI Express Retail for $379.99 and free shipping.

 

[pcslookout]

The 1070 will use less, but I don't think your power bills will make up $155 over the lifetime of your card to compensate for the huge price difference.

Edit. RS beat me to it with the clutch graphs. It really is negligible amount of power consumption difference.

Thanks as well.

 

[Arachnotronic]

Sketchy that NV's GPUs lose performance in the DX12 codepath.