Deus EX: Mankind Dividied system specs revealed

[Carfax83]

I wonder if they test using the latest AMD driver released today.
ComputerBase saying not much gains (different config) PCGH showing gains... what to do.....

The PCGH benches are actually AMDs internal benchmarks, and not PCGH..

 

[USER8000]

Looks like only now AMD has released the drivers for the DX12 version of the game:

http://www.dsogaming.com/news/amd-r...dds-dx12-support-for-deus-ex-mankind-divided/

LOL.

 

[Red Hawk]

Well another video proves my point that DX12 is a fail API. In gameplay, DX11>>>>DX12 by far on both AMD and Nvidia.
https://www.youtube.com/watch?v=TtguubFmCwo

Not sure why he's seeing those results, but on my 290X at least I definitely get better framerates walking around the streets of Prague.

 

[escrow4]

Finished this last week and deleted it. Still, the DX 12 implementation is yet more fail. If you want DX 12 build it into the engine from the ground up; you don't just shove it in in a patch.

 

[railven]

Well another video proves my point that DX12 is a fail API. In gameplay, DX11>>>>DX12 by far on both AMD and Nvidia.
https://www.youtube.com/watch?v=TtguubFmCwo

DX12 is still in the cooking process. Why I'm not sweating it. Hopefully by the time it actually matters outside of forum fights, both companies aren't seeing regression on their hardware.

To quote Danny from Game Grumps "put me back in the kiln."

EDIT: Is there a reason why this guy developed Parkinson's during the GTX 1060 in-game recording? Woof.

 

[Bacon1]

https://www.computerbase.de/2016-09/deus-ex-mankind-divided-dx12-benchmark/

computerbase sees performance regression across the board for both AMD and Nvidia with DX12 in actual gameplay in Prague level. Is everybody else seeing the same. This is why I do not trust canned in game benchmark tool as they are not reflective of actual gameplay performance.

Thanks, never liked the ROTTR benchmark either, completely different than gameplay.

Another bad DX12 port from Nixxus , hopefully they continue to work on it, ROTTR is finally better in DX12 than DX11. Can't wait for ground up Vulkan / DX12 games instead of the DX11 based patch jobs.

Has anyone checked if its using async compute? They had used it for purehair but wondering if that got scrapped on PC like it did with ROTTR and was only on console.

 

[DarkKnightDude]

Yeah seeing a huge boost in FPS for my FuryX. Even getting 80-90 FPS on streets now. Latest drivers.

 

[nFen]

Seeing a boost for my 290 ~ 10-15 fps more.

My 1060 results are fairly in-line with what Guru3d showed in terms of dx11 vs dx12 - no difference.

 

[Yakk]

Flexing the ye'old 7970 muscles with DX12.

I am now very happy playing this game.

 

[IEC]

Huge boost in Fury FPS. Also looks like 16.9.1 reduced DX11 driver overhead.

 

[Deders]

Takes much much longer to load with DX12 for me, much like Doom does with Vulcan.

 

[Red Hawk]

Takes much much longer to load with DX12 for me, much like Doom does.

That's true, load times are noticeably longer.

 

[raghu78]

Is actual gameplay performance worse or better with DX12. Can anybody who has the game test it out and let us know.

 

[Maverick177]

Is actual gameplay performance worse or better with DX12. Can anybody who has the game test it out and let us know.

Stutter mess for me with around 10% boost in performance.

 

[raghu78]

Stutter mess for me with around 10% boost in performance.

thanks for the info. I think nixxes should have released this dx12 patch when it was ready . looks rushed to me.

 

[HannooFX]

I think this patch is buggy, let's wait for the second DX12 patch

 

[Dygaza]

Is actual gameplay performance worse or better with DX12. Can anybody who has the game test it out and let us know.

Running 1080p at everything maxed. Game runs smoother than with DX11, but uses also more memory. Fury X is constantly at 3.9GB mem usage, and ram usage is around 13GB. Game needs a lot of ram with those settings. Never seen any game to use this much system memory.

 

[Red Hawk]

Stutter mess for me with around 10% boost in performance.

Yeah, it's kind of a stuttery mess for me too, even though it reports higher framerates. That combined with the agonizing load times makes DX12 mode really not worth it right now.

 

[Deders]

Is actual gameplay performance worse or better with DX12. Can anybody who has the game test it out and let us know.

I play with triple buffered vsync &75hz, with a few occasional dips below in terms of framerates.

DX12 doesn't feel quite as smooth as DX11. When the Framerates dips, it is only a few frames below DX11.

The load times for me are the killer. There is a short looped fmv of Jensen on the train when travelling between areas. I didn't realise this until I tried DX12 because in DX11, the fmv didn't have time to finish once before he got off the train. With DX12 it played through 5-7 times or more.

I'm can't remember if I was patient enough to test whether loading from a save took as long, but I didn't want to risk it.

Running 1080p at everything maxed. Game runs smoother than with DX11, but uses also more memory. Fury X is constantly at 3.9GB mem usage, and ram usage is around 13GB. Game needs a lot of ram with those settings. Never seen any game to use this much system memory.

I noticed this with Arkham Knight when I only had a 3GB card. Excess video data would be transferred over to the system ram. Before my 1080, it would reach about 12GB sysram. Now it is slightly above 8GB.

I've noticed that when in DX11 mode, Deus Ex uses about 8GB system ram and video ram readings can be over 7GB (maxed out except no msaa, 2560x1080 ultrawide)

As devs have mentioned, this is the amount that a game/app has requested, not the amount actually needed.

 

[Carfax83]

I guess I'm one of the few NVidia users wherein the game actually runs better on DX12. Just tried it, and the game is actually smoother under DX12 that it is, or should I say was under DX11. I haven't tried the DX11 renderer since updating the game admittedly, so it may be that it's been improved as well.

One thing that struck me though about the DX12 renderer, is that it uses about 1.5GB of RAM over the DX11 version. This might be what's causing stuttering for some people. If your system RAM is below 16GB, the game might start paging storage since DX12 seems to cache more data than DX11.

This could be what's causing the longer loading times as well. More data to cache, equals longer initial loading times perhaps..

 

[Deders]

Am using 16GB of ram

 

[Carfax83]

Am using 16GB of ram

Did you monitor your RAM usage in game? Mine goes well over 10GB (almost hit 11GB in fact) constantly now in Prague, but I have 32GB of RAM in my system. I'm curious to see whether you exceed the 10GB barrier.

If my theory is correct, even 16GB might not be enough to fulfill the caching requirements. Of course I expect them to tune this even more on the next patch..

 

[Deders]

I glanced at the ram graph a few times, didn't notice it go much above 8GB. Didn't play for very long in DX12 mode, but I'll have an experiment with it later on when I get a chance.

Also page file is on my Samsung 950pro.

 

[Red Hawk]

I guess I'm one of the few NVidia users wherein the game actually runs better on DX12. Just tried it, and the game is actually smoother under DX12 that it is, or should I say was under DX11. I haven't tried the DX11 renderer since updating the game admittedly, so it may be that it's been improved as well.

One thing that struck me though about the DX12 renderer, is that it uses about 1.5GB of RAM over the DX11 version. This might be what's causing stuttering for some people. If your system RAM is below 16GB, the game might start paging storage since DX12 seems to cache more data than DX11.

This could be what's causing the longer loading times as well. More data to cache, equals longer initial loading times perhaps..

Have you tested to see if CHS doesn't break shadow draw distances anymore? I didn't really look for it before, but you were criticizing the game for it. So I'm curious if the patch has fixed it in your estimation.

 

[Head1985]

Furyx vs 1070 vs 1080
https://www.youtube.com/watch?v=afA9A88hP_M
Furyx is 10% faster than 1070.