[MERGED] DirectX 11/12, OpenGL, VulkanAPI Discussion

[Cskelly]

Bought a new system. question is will DCS run with Directx 12? Found out that FSX will not, so it made me question if DCS will be able to. Thanks.

[Gladman]

I may be incorrect but I believe the new engine (2.5) is centered around DX11.

[Mustang]

Both 2.2.0 and 1.5.8 use DX11.0.

[HiJack]

Bought a new system. question is will DCS run with Directx 12? Found out that FSX will not, so it made me question if DCS will be able to. Thanks.

 

You know that you just install DX11 and the problem is solved? :music_whistling:

[Art-J]

Bought a new system. question is will DCS run with Directx 12? Found out that FSX will not, so it made me question if DCS will be able to. Thanks.

 

... and before you ask further, none of the simulators on the market use DX12 and won't do in the near future.

[BitMaster]

If they are open minded they should evaluate Vulcan and open the Doors to Mac, Linux and Windows at the same time.

 

I would, as a Producer, invest in open platform with one eye on a dedicated Linux Server for DCS. That would make a lot of sense for those who run servers. Less license fees when renting a root server, less overhead, better stability.

 

Same time, port that code to Mac and Linux for Clients as well. Open the door.

 

DX12 is nice but it glues you into the Redmond Realm = not so good

Edited December 23, 2017 by BitMaster

[Spectrum Legacy]

Vulcan is being implemented for DCS, just no idea what kind of performance boost it will provide and when it will be available. I wrote it in another thread already, but maybe since I keep typing it with 'k' as Vulkan, it wasn't obvious right away. I also wonder how it will affect the DCS's worship of singlethreaded performance. Can't wait to find out honestly, it could freshen the thing up a bit!

[cichlidfan]

The 'k' is the correct spelling.

[BitMaster]

ohhh.... O"K" then, my bad.

 

A buddy of mine runs 2 x 1070 for mining in his AMD FX at 4.8G. He has 2 new DX12 titles with mGPU and he says it blows your mind. Everything maxed out, solid high fps, ALL 8 cores used by 10-20%, he said it just flies through mGPU and solid multi-threading even on his older board and cpu. SLI hes not so happy with, works here doesnt there, he took the bridge off afaik.

[The LT]

Vulcan is being implemented for DCS, just no idea what kind of performance boost it will provide and when it will be available.

 

I pray to God that the performance boost would be along the lines of finally scaling up with multiple GPUs.

[Mr_sukebe]

I pray to God that the performance boost would be along the lines of finally scaling up with multiple GPUs.

 

From what I've read, Vulkan offloads a variety of tasks to other CPU hyperthreads. In a synthetic test, it was x3 faster than DX11. Having said that, they were also VERY specific in stating that the benefits need work by the developer to take advantage of it's capabilities. Implications being that assuming we get Vulkan, that the initial implementation might not be any better than DX11, but that as time goes on and experience is built, should lead to good things.

Can't wait..

[hansangb]

I'm OK with concentrating on Windows (and DX12). There are no gamers who buy a Mac thinking there will be a lot of support for games. The lack of advance in Mac Pro also points to why games shouldn't choose Mac. Linux is a niche within a niche solution and I don't see mainstream games coming to it. There just isn't enough for game producers to monetize.

[kunterbunt]

Most games that use direct x 12 show no performance improvement or even are slower. See battlefield 1.

I believe that DCS would profit much more from distributing the AI logic on multiple CPU cores. At least for single player scenarios. Not sure what would help for multi player.

[BeastyBaiter]

AMD GPUs gain substantial performance with DX12, Nvidia get basically nothing in most cases. Both brands see big improvements with Vulkan over DX11.

 

In regards to multiple GPU's, Vulkan doesn't support that at all last I heard. But honestly, the problem with DCS is it's single threaded graphics engine. Even getting it on 2 cores would make a massive difference. And that's what Vulkan allows in addition to a solid +30℅ fps in GPU bound situations.

[*Rage*]

I thoight ED were working on Vulkan support?

[Yurgon]

Linux is a niche within a niche solution and I don't see mainstream games coming to it. There just isn't enough for game producers to monetize.

 

DCS is the main reason I never even considered switching to Linux on my desktop. If it ran well under Linux, I might finally say goodbye to this piece of garbage Windows.

 

Other than that, it's the old chicken and egg dilemma with Linux and gaming.

[Mr_sukebe]

ED have stated that they're looking at Vulkan. No timelines of course.

Not sure why there's the comments about Linux. Vulkan as an API is for both Linux and Windows.

[tintifaxl]

If they are open minded they should evaluate Vulcan and open the Doors to Mac, Linux and Windows at the same time.

 

I would, as a Producer, invest in open platform with one eye on a dedicated Linux Server for DCS. That would make a lot of sense for those who run servers. Less license fees when renting a root server, less overhead, better stability.

 

Same time, port that code to Mac and Linux for Clients as well. Open the door.

 

DX12 is nice but it glues you into the Redmond Realm = not so good

 

DX is not only about graphics. DX offers a lot more: sound, network and input devices. All of these would have to be ported, too.

[BitMaster]

DCS is the main reason I never even considered switching to Linux on my desktop. If it ran well under Linux, I might finally say goodbye to this piece of garbage Windows.

 

Other than that, it's the old chicken and egg dilemma with Linux and gaming.

 

+1

[BitMaster]

DX is not only about graphics. DX offers a lot more: sound, network and input devices. All of these would have to be ported, too.

 

I play many games on Mac, some still on Linux. ASIO is there for sound, and controller files are not that complicated for Mac, and cant be that hard for Linux. I dont see an obstacle there.

 

YOu could boot a Linux just for gaming, imagine ;)

[SkateZilla]

DCS World Currently uses DirectX 11_0

 

With a planned transition to VulkanAPI

[BitMaster]

DCS World Currently uses DirectX 11_0

 

With a planned transition to VulkanAPI

 

 

That's what we actually wanted to hear :thumbup:

 

Why took so long ?

[The LT]

I play many games on Mac, some still on Linux. ASIO is there for sound, and controller files are not that complicated for Mac, and cant be that hard for Linux. I dont see an obstacle there.

 

YOu could boot a Linux just for gaming, imagine ;)

 

ASIO is a proprietary API for semi-pro and pro cards. Main audio api for Linux is ALSA.

[BitMaster]

ASIO is a proprietary API for semi-pro and pro cards. Main audio api for Linux is ALSA.

 

correct ! all those abbreviations...or was it ALDI ?

[firmek]

That's what we actually wanted to hear :thumbup:

 

Why took so long ?

:) We can assume it'll take really long to migrate to Vulcan, without any guarantee of a performance boost. It's a lot of work, requires to rewrite the whole rendering engine at least. In reality probably much more especially in an quite old application as DCS there is no pure rendering engine - "business objects"/logic separation. On top of that it's a totally different API which requires a different "mind-set" when programming.

 

Obviously I don't know the details but to set the expectations I would expect even more work than the whole 2.5 merge migration.

Not being negative but just stating the facts as it seems according to some posts we could get an impression that introducing Vulcan is guaranteed to triple the performance and will happen in DCS in 3-6 months :). Not even mentioning that just by going to Vulcan will magically make DCS to be multi-platform.

Take EA/DICE and BF1 as an example. With the budget of AAA title their DX12 renderer is still experimental and yields a worse performance than the old, DX11 one with no noticeable visual difference.