Deus EX: Mankind Dividied system specs revealed

[2is]

Any benefit to running this game in DX12 with a 980Ti?

 

[tviceman]

Did not know that. I thought all the UWP games were DX12. Anyway, this patch is a HUGE improvement to Deus Ex MD. If anyone was holding off until the game was patched up before buying it, now is the time to buy it.. The difference in performance between launch state (which was DX11 only) and the game now is unbelievable.

Deus Ex MD as it stands right now (and other games like Gears 4, AotS), really shows the potential of DX12 when done properly.. The resource management is just so much better under DX12.

Interesting! Too bad there isn't a built in benchmark to test dx11 vs. dx12.

 

[Erenhardt]

Did not know that. I thought all the UWP games were DX12. Anyway, this patch is a HUGE improvement to Deus Ex MD. If anyone was holding off until the game was patched up before buying it, now is the time to buy it.. The difference in performance between launch state (which was DX11 only) and the game now is unbelievable.

Deus Ex MD as it stands right now (and other games like Gears 4, AotS), really shows the potential of DX12 when done properly.. The resource management is just so much better under DX12.

Any benchmarks to show what you are talking about?

 

[moonbogg]

So does this game use multi GPU yet?

 

[Bacon1]

So does this game use multi GPU yet?

Yes they've had DX11 support for a while and added MGPU a week or two ago for DX12, though it's still in testing / beta.

 

[Headfoot]

ill buy it now, I was waiting for dx12 mGPU to buy. Loved the first one

 

[Carfax83]

Interesting! Too bad there isn't a built in benchmark to test dx11 vs. dx12.

I think there is a built in benchmark, though I don't know how useful it would be to determine the worth of DX12 vs DX11. The benchmark is very GPU bound, while the game itself uses the CPU a lot more, especially in Prague. I'll run some quick benches though and post the results..

 

[Carfax83]

Any benefit to running this game in DX12 with a 980Ti?

DX12 performance has improved substantially since the latest patch, so you could try giving it a go. Most websites test the inbuilt benchmark though which is crap and not really indicative of actual game performance..

 

[Carfax83]

As I thought, the in game benchmark is absolutely useless. DX11 and DX12 are about even here. Like I said before, the benchmark is way more GPU bound than anything in the actual game itself.

From my experience, DX12 runs noticeably better now than DX11 after this latest patch. Also, I'm getting 60 FPS almost constantly in the actual game using these settings, unlike the misleading benchmark:

DX11:

DX12:

 

[Bacon1]

As I thought, the in game benchmark is absolutely useless. DX11 and DX12 are about even here. Like I said before, the benchmark is way more GPU bound than anything in the actual game itself.

From my experience, DX12 runs noticeably better now than DX11 after this latest patch. Also, I'm getting 60 FPS almost constantly in the actual game using these settings, unlike the misleading benchmark:

DX11:

DX12:

Can you test Ultra vs Low textures while you are at it?

 

[Azix]

As I thought, the in game benchmark is absolutely useless. DX11 and DX12 are about even here. Like I said before, the benchmark is way more GPU bound than anything in the actual game itself.

From my experience, DX12 runs noticeably better now than DX11 after this latest patch. Also, I'm getting 60 FPS almost constantly in the actual game using these settings, unlike the misleading benchmark:

DX11:

DX12:

then just run thru a segment of gameplay in dx11 then the same in dx12. presentmon or some other app can monitor the fps. No need for guessing its better fps overall. Could just be smoother. The benchmark is close to worst case scenario so you won't get 60 fps but it could still be showing an accurate view of dx11 vs dx12 performance for your system.

 

[Carfax83]

Can you test Ultra vs Low textures while you are at it?

This is one with DX12 with the same settings as before, but with the textures on low quality. So using low quality textures increases performance by 3.3% compared to ultra..

 

[Bacon1]

This is one with DX12 with the same settings as before, but with the textures on low quality. So using low quality textures increases performance by 3.3% compared to ultra..

Thanks, what resolution was that at?

 

[Carfax83]

then just run thru a segment of gameplay in dx11 then the same in dx12. presentmon or some other app can monitor the fps. No need for guessing its better fps overall. Could just be smoother. The benchmark is close to worst case scenario so you won't get 60 fps but it could still be showing an accurate view of dx11 vs dx12 performance for your system.

DX12 is definitely smoother, which implies more consistent frame latency. I've tested this many times myself and I don't really need testing software to know that. Deus Ex MD is really a great example of a game that can potentially benefit from DX12, because the game is so incredibly dense with a high amount of draw calls and loads of high quality textures. With DX12, my VRAM readout is always well over 7GB (about 7.7GB) with ultra quality textures enabled, and my RAM usage is over 8GB for the working set, and 11GB+ for the commit.

WIth DX11, my memory usage is noticeably lower, VRAM slightly below 7GB, and the working set is 5GB+ and commit around 8GB or so. This leads to the game being noticeably less smoother in DX11..

CPU usage is about the same between them I think, but with DX12 more of the CPU is being used evenly, whereas with DX11 core0 gets a disproportionate amount of the load.

 

[Carfax83]

Thanks, what resolution was that at?

1440p.

 

[2is]

DX12 is definitely smoother, which implies more consistent frame latency. I've tested this many times myself and I don't really need testing software to know that. Deus Ex MD is really a great example of a game that can potentially benefit from DX12, because the game is so incredibly dense with a high amount of draw calls and loads of high quality textures. With DX12, my VRAM readout is always well over 7GB (about 7.7GB) with ultra quality textures enabled, and my RAM usage is over 8GB for the working set, and 11GB+ for the commit.

WIth DX11, my memory usage is noticeably lower, VRAM slightly below 7GB, and the working set is 5GB+ and commit around 8GB or so. This leads to the game being noticeably less smoother in DX11..

CPU usage is about the same between them I think, but with DX12 more of the CPU is being used evenly, whereas with DX11 core0 gets a disproportionate amount of the load.

This may be the case for Pascal cards, Maxwell on the other hand more often then not has worse performance in DX12. By more often then not, I mean very close to 100% of the time.

 

[Carfax83]

This may be the case for Pascal cards, Maxwell on the other hand more often then not has worse performance in DX12. By more often then not, I mean very close to 100% of the time.

I suppose it depends. If the game is CPU bound, then there's no reason why Maxwell can't deliver a performance increase with DX12, as that's where DX12 nets most of the performance gains from anyway. But when it's GPU bound, then yeah I agree, that Maxwell is likely to have more of a problem with DX12 as it lacks asynchronous compute..

 

[Bacon1]

Computerbase did a "final test" for DX12:

https://www.computerbase.de/2016-10/deus-ex-dx12-final-benchmark/

 

[Bacon1]

Also they announced its coming to Linux.

https://twitter.com/feralgames/status/789392785073770496

They haven't announced Vulkan vs OpenGL though:

https://twitter.com/feralgames/status/789454199654277121

 

[AtenRa]

Computerbase did a "final test" for DX12:

https://www.computerbase.de/2016-10/deus-ex-dx12-final-benchmark/

hmm, still needs optimization. FX8370 same fps in DX-11 vs DX-12 with RX 480, when GTX 1060 is used DX-11 is faster than DX-12.

 

[Red Hawk]

AMD fixed the issue with their drivers where CPUs that didn't support the popcnt instruction set, such as the Q6600 in my brother's PC, couldn't run DX12 renderers at all. Now they can, so I can pose the question: How much of a difference can DX12 make on an aging PC with a stock Core 2 Quad Q6600, Radeon R9 270X, and 8 GB of 800 MHz DDR2 RAM?

Benchmark results at 1440x900, High preset, Exclusive Fullscreen on, Triple Buffering on:

DX11:
Average: 26.0
Minimum: 8.6
Maximum: 40.0

DX12:
Average: 32.5
Minimum: 8.6
Maximum: 47.8

Answer: A fair bit! Though it is concerning that the minimum doesn't budge.

 

[Bacon1]

AMD fixed the issue with their drivers where CPUs that didn't support the popcnt instruction set, such as the Q6600 in my brother's PC, couldn't run DX12 renderers at all. Now they can, so I can pose the question: How much of a difference can DX12 make on an aging PC with a stock Core 2 Quad Q6600, Radeon R9 270X, and 8 GB of 800 MHz DDR2 RAM?

Benchmark results at 1440x900, High preset, Exclusive Fullscreen on, Triple Buffering on:

DX11:
Average: 26.0
Minimum: 8.6
Maximum: 40.0

DX12:
Average: 32.5
Minimum: 8.6
Maximum: 47.8

Answer: A fair bit! Though it is concerning that the minimum doesn't budge.

Wow that's a 25% increase in FPS. Makes you able to hit locked 30 pretty well.

 

[Red Hawk]

Wow that's a 25% increase in FPS. Makes you able to hit locked 30 pretty well.

Unfortunately, actual gameplay didn't back those numbers up. Using Steam's frame rate counter, I got around 17-20 FPS walking around Prague in DX12, while in DX11 I was getting 20-24 FPS. Clear advantage for the DX11 renderer. Darn.

 

[Carfax83]

Computerbase did a "final test" for DX12:

https://www.computerbase.de/2016-10/deus-ex-dx12-final-benchmark/

I think with a few more performance patches, Deus Ex MD might just become one of the premier DX12 titles, along with AotS and Gears 4.

 

[Bacon1]

Unfortunately, actual gameplay didn't back those numbers up. Using Steam's frame rate counter, I got around 17-20 FPS walking around Prague in DX12, while in DX11 I was getting 20-24 FPS. Clear advantage for the DX11 renderer. Darn.

Hmm thats very odd, what was CPU / GPU usage in both?