[guru3d] Total War: Warhammer DX12 benched

[biostud]

On my setup

DX11/DX12:
Ultra: 47.8/28.1
High: 65.9/36.9
Medium: 79.1/44.2
Low: 97.0/56.4

Not really surprising since the 7990 is not officially supported in DX12.

 

[3DVagabond]

On my setup

DX11/DX12:
Ultra: 47.8/28.1
High: 65.9/36.9
Medium: 79.1/44.2
Low: 97.0/56.4

Not really surprising since the 7990 is not officially supported in DX12.

Is that because of lack of crossfire support?

 

[Bacon1]

Is that because of lack of crossfire support?

There is CFX support AFAIK, gamegpu showed scaling, but it was only noticeable at 1440 a small bit, and then much better @ 4k.

 

[SPBHM]

On my setup

DX11/DX12:
Ultra: 47.8/28.1
High: 65.9/36.9
Medium: 79.1/44.2
Low: 97.0/56.4

Not really surprising since the 7990 is not officially supported in DX12.

can you test both with CF disabled? maybe it's just CF not working on DX12

 

[biostud]

can you test both with CF disabled? maybe it's just CF not working on DX12

DX11 no CF
Ultra:27.8
High: 36.1
Medium: 44.2
Low: 56.9

And the exact same numbers in DX12, so DX12 CF is not working for 7990, and maybe it simply uses DX11 since the card is not supported in DX12.

 

[Yakk]

Crossfire is not supported in the DX12 version (yet?) sadly.

However I prefer playing with only one of my 280x cards in DX12 over 2x 280x in DX11. Overall gameplay in DX12 feels much better, 7900 series are supported, but with only 2 ACE units, they get more modest improvements compared to a 390x with 8 ACE units. I get a 3 to 5 FPS improvement from DX11 to DX12 on my 280x.

 

[Zstream]

How are you guys doing benchmarks?

 

[Yakk]

How are you guys doing benchmarks?

New benchmark ingame added in last patch with DX12.

 

[biostud]

How are you guys doing benchmarks?

Go to graphics-> advanced -> there is now a benchmark button.

 

[biostud]

Crossfire is not supported in the DX12 version (yet?) sadly.

However I prefer playing with only one of my 280x cards in DX12 over 2x 280x in DX11. Overall gameplay in DX12 feels much better, 7900 series are supported, but with only 2 ACE units, they get more modest improvements compared to a 390x with 8 ACE units. I get a 3 to 5 FPS improvement from DX11 to DX12 on my 280x.

It could be supported by newer videocards, but someone else than me would have to confirm it.

I get less than 1 fps difference between dx11 and dx12 without cf.

 

[biostud]

So for me cf scaling in dx11

Ultra 70%
High 83%
Medium 79%
Low 70%

 

[Yakk]

It could be supported by newer videocards, but someone else than me would have to confirm it.

I get less than 1 fps difference between dx11 and dx12 without cf.

Hmm I'm running Asus DC2 280x cards which run stock at 1070mhz, maybe a difference in card speed increases the difference. I'll run it again OC'd and see. Benchmark I get about a 3fps difference, ingame it is close to 5 overall. The benchmark run doesn't include very large scale battles unfortunately.

 

[biostud]

Hmm I'm running Asus DC2 280x cards which run stock at 1070mhz, maybe a difference in card speed increases the difference. I'll run it again OC'd and see. Benchmark I get about a 3fps difference, ingame it is close to 5 overall. The benchmark run doesn't include very large scale battles unfortunately.

What CPU do you have, could be a difference between hex vs quad.

 

[Yakk]

What CPU do you have, could be a difference between hex vs quad.

I just use an old i7 920 @4.2ghz with HT on. One thing I noticed, is under DX11 the CPU thread #8 is pegged at 100% (unless you do the turn off/on trick which sometimes causes game crashes), but under DX12 that #8 thread hovers around 88-97%. So a bottleneck on that main thread may have been removed improving the overall game performance.

 

[Zstream]

Cool - Asus Strix 280 and an E3-1231 v3

I get 43FPS

Everything is Ultra
DoF - Off
Screen Space reflections - Off
AA - FXAA
TF - Anti 16x
Unit Size - Large
SSAO
Distortion effects
Proxy fading
Blood effects

 

[biostud]

I just use an old i7 920 @4.2ghz with HT on. One thing I noticed, is under DX11 the CPU thread #8 is pegged at 100% (unless you do the turn off/on trick which sometimes causes game crashes), but under DX12 that #8 thread hovers around 88-97%. So a bottleneck on that main thread may have been removed improving the overall game performance.

That could definitely explain the difference. You often see improvements for AMD and DX12 on "weak" CPUs.

 

[biostud]

Cool - Asus Strix 280 and an E3-1231 v3

I get 43FPS

Everything is Ultra
DoF - Off
Screen Space reflections - Off
AA - FXAA
TF - Anti 16x
Unit Size - Large
SSAO
Distortion effects
Proxy fading
Blood effects

If you just choose the presets it will be easier to compare, I guess you run on the ultra preset? What resolution?

All mine are 2560x1440

 

[Zstream]

If you just choose the presets it will be easier to compare, I guess you run on the ultra preset? What resolution?

All mine are 2560x1440

1920 * 1080

Yeah, it's all preset - I didn't touch a thing when it was installed.

 

[Zstream]

Cool - Asus Strix 280 and an E3-1231 v3

I get 43FPS

Everything is Ultra
DoF - Off
Screen Space reflections - Off
AA - FXAA
TF - Anti 16x
Unit Size - Large
SSAO
Distortion effects
Proxy fading
Blood effects

New 480 (default clocks) 78FPS - A near 180% increase

 

[AtenRa]

New 480 (default clocks) 78FPS - A near 180% increase

You mean 81% increase

 

[antihelten]

New 480 (default clocks) 78FPS - A near 180% increase

That's not really how percentages work (it's an 80% increase not 180%).

 

[Zstream]

That's not really how percentages work (it's an 80% increase not 180%).

It's a total of 180% of my 280, there.

 

[antihelten]

It's a total of 180% of my 280, there.

correct