DirecxtX 12 for ArmA 3?

[calin_banc 19]

We've got this from one of the status updates on Day Z:

 

 

internally we've been observing outstanding client side performance (frame rate) with the new renderer technology. Paired with the performance increase of the New UI, and iteration upon the inventory and UI skinning - we're *hoping* to see .60 be a hell of an update.

 

Corroborating with what they've said about no real benefits from Dx12 and no plans for deep engine optimizations, then all this talk about DX12 and Vulkan is a moot point. Instead of the next add-on, personally I would have loved to see Bohemia working on ArmA 4 on the new engine. New graphical updates and such are great (hopefully some meaningful UI and character animations/interactions/controls), but still, playing that subpar fps...

[I give up 152]

For those who still believe in DX12 miracle in matters of performance, I suggest taking a look at this:

http://www.techspot.com/review/1081-dx11-vs-dx12-ashes/

I cant see the widely claimed improvements about performance, and you?

[calin_banc 19]

You need to have another look at that article.

[Vasily.B 529]

You need to have another look at that article.

I also checked the site - on 3rd pages is the FPS table - on some setups there is 3-5 FPS more.... WOW thats all!?

[_.Dardo._ 15]

I also checked the site - on 3rd pages is the FPS table - on some setups there is 3-5 FPS more.... WOW thats all!?

If drivers and hardware are not built for working with DX12,obviously we won't have any kind of performance increases. Unfortunately,we'll have to wait to see the muscles of dx12 .

[Vasily.B 529]

If drivers and hardware are not built for working with DX12,obviously we won't have any kind of performance increases. Unfortunately,we'll have to wait to see the muscles of dx12 .

And install Win10 - thanks a lot, i'd rather wait for vulcan.

[_.Dardo._ 15]

And install Win10 - thanks a lot, i'd rather wait for vulcan.

...if it's not going to end like mantle... Lol

[en3x 209]

If we look back at dx11 that was windy start too.We can learn that every new technology needs time for developers

and vendors to optimize and learn their way around.

[linuxmaster9 101]

...if it's not going to end like mantle... Lol

The likelihood of it ending like Mantle is minuscule. Mantle is from AMD alone. Vulkan is from Khronos Group. The same group behind OpenGL. Also Vulkan is backed by somewhere in the realm of 50 companies including EA, Sony, Microsoft, Unreal, and a host of others. Vulkan has industry backing unlike Mantle.

[_.Dardo._ 15]

The likelihood of it ending like Mantle is minuscule. Mantle is from AMD alone. Vulkan is from Khronos Group. The same group behind OpenGL. Also Vulkan is backed by somewhere in the realm of 50 companies including EA, Sony, Microsoft, Unreal, and a host of others. Vulkan has industry backing unlike Mantle.

Didn't know this.

Inviato dal mio LG-D855 utilizzando Tapatalk

[calin_banc 19]

I also checked the site - on 3rd pages is the FPS table - on some setups there is 3-5 FPS more.... WOW thats all!?

http://imgur.com/ernOW8s

http://imgur.com/xmpufQB

 

http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.pcgameshardware.de%2FAMD-Radeon-Grafikkarte-255597%2FSpecials%2FAMD-Mantle-Leistungsangaben-Treiber-Battlefield-4-Patch-unterwegs-1107277%2F&anno=2&sandbox=1

http://www.pcgameshardware.de/AMD-Radeon-Grafikkarte-255597/Specials/AMD-Mantle-Test-Battlefield-4-1107754/

http://www.tomshardware.com/reviews/amd-mantle-performance-benchmark,3860-4.html

http://www.techspot.com/review/793-thief-battlefield-4-mantle-performance/page3.html

 

 

That BF4 screnshots in game is mine. So ~ 23% bonus with a low level API - Mantle. Please look at those graphs down on the left. Mantle allows, under the same settings, to have a bigger performance, ALTHOUGH neither the CPU or the GPU is limiting each under dx11.(2? I think BF uses?) - those two lines are basically one on top of the other.

 

2nd picture is with different results picked up from the web with Mantle vs. DX11.x as well just to show how low level apis can boos performance. You also see those DX12 results, look again!

 

You guys need to understand that it depends where and how you test a game (after all, you can be GPU limited even in ARMA if you really play with the settings). Also it matters how well you code and really put effort into making your engine actually benefit from such an approach. Again, ArmA 3 my be a dx11 game, but it doesn't use tessellation (as far as I know) and it doesn't use tilled resources (dx11.2) which should be beneficial since streaming large amount of textures and poor terrain textures are some of the game's problems. Ergo, just because a game is using DX12, it doesn't mean it's very good at it.

[St. Jimmy 272]

This is the quote why I'd like to see some video of usages and all that:

 

 

It's worth noting that our GPU testing was conducted with the ultra-fast Core i7-6700K processor, which is why DX12 didn't show vast improvements over DX11 -- there were loads of processing power available. That said, when we turned to test different CPUs, even with a Core i3 processor we had to reduce the graphics settings drastically to alleviate the GPU bottleneck.

 

With a weaker CPU it seems to move the work more to the GPU. Kind of like it just improves dual cores but with higher ens stuff there's no improvement? i3 you get a lot of GPU usage (too much for one card) but with i7 you don't get?

[I give up 152]

The likelihood of it ending like Mantle is minuscule. Mantle is from AMD alone. Vulkan is from Khronos Group. The same group behind OpenGL. Also Vulkan is backed by somewhere in the realm of 50 companies including EA, Sony, Microsoft, Unreal, and a host of others. Vulkan has industry backing unlike Mantle.

And Nvidia.

ARMA 3 is powered by Nvidia (graphically speaking) I can't find a reason to not look at it. For AMD anything is better than DX11..

[calin_banc 19]

This is the quote why I'd like to see some video of usages and all that:

 

 

 

With a weaker CPU it seems to move the work more to the GPU. Kind of like it just improves dual cores but with higher ens stuff there's no improvement? i3 you get a lot of GPU usage (too much for one card) but with i7 you don't get?

 

 

It doesn't move anything from the CPU, DX12/Vulkan/Mantle/whatever can use that CPU more efficiently.  Yes, you do see a boost with an i7 http://www.techspot.com/articles-info/1081/bench/CPU_06.png

[linuxmaster9 101]

And Nvidia.

ARMA 3 is powered by Nvidia (graphically speaking) I can't find a reason to not look at it. For AMD anything is better than DX11..

True. I was doing many things when I typed that out so I was pulling the immediate names I could recall. Intel, AMD, Nvidia, LG, Apple, ARM, Sony, Samsung, Epic Games, Imagination, Nokia, Qualcomm, Lucasfilm, NEC, Panasonic, Oxide, Oculus VR, Pixar, Red Hat, Unity, Toshiba, Texas Instruments, Valve, Wargaming.net, vmware, Mediatek, Matrox, Mozilla, IBM, Codeplay, Adobe, Amazon, Altera, Blizzard, Broadcom, Canonical, Google, Futuremark, and Gaijin, etc are all members of the Khronos Group board of Promoters. There are over 100 Members worldwide.

All of them back Vulkan.

 

 

Who is backing DX12? Microsoft and XBox (split from Microsoft).

[batto 17]

These DirectX benchmarks are mostly pointless. It's the game and the hardware that is responsible for moar framez. The rendering API and drivers should have as little overhead as possible. That review misses the main newsworthy of DirectX 12 (and Vulkan) which is the low-level interface, which should greatly reduce this overhead, but the game needs to be vastly redesigned for it first. Hopefully these DirectX benchmarks will become even more pointless in future.

 

Also, AMD donated Mantle to Khronos, which, along with major players in the industry (except Microsoft), turned it into Vulkan. So it's better to wait for its release and forget about Mantle.

[St. Jimmy 272]

It doesn't move anything from the CPU, DX12/Vulkan/Mantle/whatever can use that CPU more efficiently.  Yes, you do see a boost with an i7 http://www.techspot.com/articles-info/1081/bench/CPU_06.png

Yeah and the ultrasuperhyper graphic setup seems to confuse. That's why I don't like those kind of reviews because they don't show enough information what's going on. Everyone naturally watches the highest graphic setting and don't see improvement so they're like "oh it's nothing" when the bottleneck was in the GPU in that but DX12 is suppose to improve CPU so it can do more of the simulation. There's nothing that says how was the GPU load in the highest graphic settings though thinking when comparing it to the medium settings the GPU was the bottleneck there.

 

AMD GPUs will get nice improvement I've read that before. Nvidia not much but they've said they weren't ready yet. AMD at least has a head start because of Mantle.

[I give up 152]

Well, like I have said several times, DX12 is a marketing campaign to force people to get Windows 10.
Last 6 months I have been using Windows 10, mainly because DX12. I've tried several applications under DX12 and honestly none of them justify the spyware that I have been forced to tolerate.
After killing cortana and her friends, after killing all forms (that I could find) of spyware provided by Windows 10, everytime time I left my machine alone I have found it communicating with Microsoft.
DX12 is not enough, I need much more to make me lose my privacy.

I am back to Windows 7, but it's amazing what Microsoft can do to force Windows 10.
In the past I could get all updates for Windows 7 in a matter of minutes, now it takes 3 days (or more). Just amazing.

 

Vulkan is independent, Vulkan does not have such things, that's reason enough to support it. Even more when there is benefit for everyone.

[linuxmaster9 101]

Yeah and the ultrasuperhyper graphic setup seems to confuse. That's why I don't like those kind of reviews because they don't show enough information what's going on. Everyone naturally watches the highest graphic setting and don't see improvement so they're like "oh it's nothing" when the bottleneck was in the GPU in that but DX12 is suppose to improve CPU so it can do more of the simulation. There's nothing that says how was the GPU load in the highest graphic settings though thinking when comparing it to the medium settings the GPU was the bottleneck there.

 

AMD GPUs will get nice improvement I've read that before. Nvidia not much but they've said they weren't ready yet. AMD at least has a head start because of Mantle.

Also, fun fact, DX12 imitates Mantle because it does not contain Mantle code. Microshaft decided not to take AMD up on their offer of Mantle. Thought they could do it better themselves. Obviously, not.

[_.Dardo._ 15]

Guys ,I'll say it again,you can NOT talk about benchmarks with DirectX12 if hardware hasn't been optimised for working with these libraries. I can't tell if they are better or worse than Vulkan API. We will know it only around the end of 2016, or later.. When optimised hardware will be out.

[PuFu 4600]

Also, fun fact, DX12 imitates Mantle because it does not contain Mantle code. Microshaft decided not to take AMD up on their offer of Mantle. Thought they could do it better themselves. Obviously, not.

i understand you want to play A3 on something else but windows, but if mantle was such a fucking great success (which it wasn't) why the fuck didn't AMD release the SDK and instead simply pulled the plug on it...?

Yes vulkan backers (not partners mind you) are numerous indeed, and i have already seen blogs from nvidia and amd stating that their drivers would be ready on the release date. That being said, i also see companies such as Apple, which is indeed using OpenGL, but on the same time have spent an enourmous amount of money on R&D for their own Metal implementation. And this is just an example.

 

Since there is NOT a single 100% DX12 GFX card available out there, and since Vulkan is not released in any sort of form, i really don't get how you can make a comparison, or why do you have such a blind faith in Vulkan without a single piece of evidence.  

[calin_banc 19]

Yeah and the ultrasuperhyper graphic setup seems to confuse. That's why I don't like those kind of reviews because they don't show enough information what's going on. Everyone naturally watches the highest graphic setting and don't see improvement so they're like "oh it's nothing" when the bottleneck was in the GPU in that but DX12 is suppose to improve CPU so it can do more of the simulation. There's nothing that says how was the GPU load in the highest graphic settings though thinking when comparing it to the medium settings the GPU was the bottleneck there.

 

AMD GPUs will get nice improvement I've read that before. Nvidia not much but they've said they weren't ready yet. AMD at least has a head start because of Mantle.

 

 

You need to know where to look for those differences, if they are testing what they suppose to or not. Here, let me give you an example in ArmA 3 how can I go pretty easily from a GPU bound situation to a something else bound situation - I'm not going to say it's the CPU or whatnot, it doesn't matter:  http://imgur.com/Vin5ZNt

 

Of course if you just test under the first settings, the gains under a more efficient way of doing things, whatever that may be, will be minimal. The limiting factor is the GPU. Now go to the 2nd crop and 3rd where the GPU is not anymore the bottleneck, but something else.

 

1st crop is at 5280x1050, 8xAA, 133% resolution (so 33% above 5280*1050), 2k for object and view distance, ultra terrain, low details objects. The other important settings are maxed. Of course, under these settings a r290 even with overclock is not capable of delivering a stellar performance.  At this point you're in the situation of the above DX12 test while using max settings - or at least ones that are GPU bound.

 

2nd, 3rd crop are in situations where you're not GPU bound, but something else is keeping your frame rate down. This is the situation on that DX12 where you have medium settings. Not to be understood that DX12 will fix ArmA 3 performance issues, but rather you're starting to test what you suppose to in DX12, Vulkan, etc., in that AoS benchmark.

 

4th crop is where something else is gone/paused în ArmA 3, so the GPU can be used closer to its full potential, due to more time available on the main core(s) of the CPU.

 

Here is the full image if interested: http://imgur.com/SFXfDKL

 

 I have been using Windows 10, mainly because DX12. I've tried several applications under DX12 and honestly none of them justify the spyware that I have been forced to tolerate.

 

 

And what exactly have you been testing? Besides 3dmark test and AoS, both in their infancies, there is no other application for dx12. Just by having DX12 in your system does not do miracles, just as having rocket fuel in the back of my car, doesn't allow me to go many Machs.

 

 why the fuck didn't AMD release the SDK and instead simply pulled the plug on it...?

 

Because it wasn't finished. At a point they had around 70-80 developers working with Mantle, said nVIDIA is welcomed to it and that they will release it once it's done: it reached beta, if I'm not mistaken. In the mean time DX12 and Vulkan surfaced, so there is no point anymore for Mantle, more so when nVIDIA was vocal not going Mantle. :)

[linuxmaster9 101]

i understand you want to play A3 on something else but windows, but if mantle was such a fucking great success (which it wasn't) why the fuck didn't AMD release the SDK and instead simply pulled the plug on it...?

Yes vulkan backers (not partners mind you) are numerous indeed, and i have already seen blogs from nvidia and amd stating that their drivers would be ready on the release date. That being said, i also see companies such as Apple, which is indeed using OpenGL, but on the same time have spent an enourmous amount of money on R&D for their own Metal implementation. And this is just an example.

 

Since there is NOT a single 100% DX12 GFX card available out there, and since Vulkan is not released in any sort of form, i really don't get how you can make a comparison, or why do you have such a blind faith in Vulkan without a single piece of evidence.  

I find it funny that you dont think there is any 100% DX12 hardware out there. AMd Fury series for example. It is DX12 and OpenGL 4.5. Oh yeah, a Backer is a paying member of the Khronos Group......I'd call that a partner.

 

AMD did release the SDKs. Also, keep in mind that AMD was short on cash (still is) and could not afford to keep developing an API that was only going to hit half the market. I have such huge faith in Vulkan not because I am blind but because I have used OpenGL before.

I also dont understand where you are getting the whole thing about playing the game not on Windows. I am all for playing on Windows. Just not on Windows 10. I would love to go full Linux but until my library is large enough to change my boot loader to boot to the dedicated Linux SSD first, I will continue on Windows 7 not effin 10 (piece of shit OS).

 

And based on the development demos from Intel, Imagination Tech, and Nvidia, I would have to say that Vulkan is looking sexy. Vulkan is not released to the public as of yet.......but, Devs do have access to it if they are a member of the Group. Example being how Source 2 uses Vulkan and not DX12.....yet Vulkan ain out to the public yet and Valve already has working Source2 engine just waiting for the official release date to roll it out.

 

I personally dont see how you can have such blind faith in DX12 which has not shown any real benefit. DX12 does not have the proof of concept that Vulkan has (Mantle). DX12 is re-written DX11 code for Lowlevel API access. Mantle already handles it. It was also already shown in BF4 to work very well. Given how AMD gave the project to Khronos Group who is known for the OpenSource platform (SDK anyone?) OpenGL, I would have to say that the future of Vulkan is great.

[_.Dardo._ 15]

I find it funny that you dont think there is any 100% DX12 hardware out there. AMd Fury series for example. It is DX12 and OpenGL 4.5. Oh yeah, a Backer is a paying member of the Khronos Group......I'd call that a partner.

AMD did release the SDKs. Also, keep in mind that AMD was short on cash (still is) and could not afford to keep developing an API that was only going to hit half the market. I have such huge faith in Vulkan not because I am blind but because I have used OpenGL before.

I also dont understand where you are getting the whole thing about playing the game not on Windows. I am all for playing on Windows. Just not on Windows 10. I would love to go full Linux but until my library is large enough to change my boot loader to boot to the dedicated Linux SSD first, I will continue on Windows 7 not effin 10 (piece of shit OS).

And based on the development demos from Intel, Imagination Tech, and Nvidia, I would have to say that Vulkan is looking sexy. Vulkan is not released to the public as of yet.......but, Devs do have access to it if they are a member of the Group. Example being how Source 2 uses Vulkan and not DX12.....yet Vulkan ain out to the public yet and Valve already has working Source2 engine just waiting for the official release date to roll it out.

I personally dont see how you can have such blind faith in DX12 which has not shown any real benefit. DX12 does not have the proof of concept that Vulkan has (Mantle). DX12 is re-written DX11 code for Lowlevel API access. Mantle already handles it. It was also already shown in BF4 to work very well. Given how AMD gave the project to Khronos Group who is known for the OpenSource platform (SDK anyone?) OpenGL, I would have to say that the future of Vulkan is great.

Please check your sources.. Fury does not fully support dx12..

Inviato dal mio LG-D855 utilizzando Tapatalk

[linuxmaster9 101]

Since some people seem to have a bad idea of what the Khronos Group is, let me settle this once and for all:

 

The Khronos Group

Who is it, What is it

 

The Khronos Group is a non-profit, member-funded industry consortium founded in the year 2000 and based out of Beaverton, Oregon. The consortium is focused on the creation of royalty-free, open standards for parallel computing, graphics and dynamic media on a wide variety of platforms and devices. All Khronos members are able to contribute to the development of Khronos API specifications, are empowered to vote at various stages before public deployment, and are able to accelerate the delivery of their cutting-edge 3D platforms and applications through early access to specification drafts and conformance tests. (Source: https://www.khronos.org/about)

 

The Khronos Group was founded by a number of leading media-centric companies, including 3Dlabs, ATI/AMD, Discreet, Evans & Sutherland, Intel, Nvidia, SGI and Sun Microsystems.

 

There are 5 levels of Memberships: Implementers, Adopters, Academic Contributors, Contributors, and Promoters.

 

Implementers may create and deliver a product using the publicly released specifications, but cannot use the trademarks.

Adopters complete the conformance testing procedures and sign a royalty-free license to use the trademarks on their products.

Academic Contributors have full API working group participation but no voting rights.

Contributors have full API working group participation and voting rights and generous marketing benefits.

Promoters act as the "Board of Directors" to set the direction of the Group with the final specification ratification voting rights.

 

 

 

The Khronos Group now has roughly 120 member companies, over 30 adopters, and 24 conforming members.

 

The following is a list of the Working Groups or projects:

 

• COLLADA
• EGL
• OpenCL
• OpenGL
• OpenGL ES
• OpenGL SC
• OpenKCam
• OpenKODE
• OpenMAX
• OpenML
• OpenSL ES
• OpenVG
• OpenVX
• OpenWF
• SPIR
• StreamInput
• Vulkan
• WebCL
• WebGL

 

The following is a list of the Promoters member companies:

 

• AMD/ATI
• Apple inc
• ARM Holdings
• Epic Games
• Imagination Technologies
• Intel Corporation
• Nokia
• Nvidia
• Qualcomm
• Samsung Electronics
• Sony Computer Entertainment

 

 

The following is a list of their Contributors:

 

• Adobe
• Amazon.com
• Blizzard Entertainment inc.
• Codeplay
• Ericsson
• Google
• Huawei
• IBM
• LG
• Lucasfilm Ltd
• Matrox Graphics
• Microsoft Corporation
• Mozilla
• Nintendo
• Oculus VR
• Panasonic
• Pixar
• Red Hat
• Renesas Electronics
• Synopsys
• Texas instruments
• Unity Technologies
• Valve Corporation
• VIA Alliance Semiconductor
• VMware

 

 

So, based on the details I have provided, When I mention they are Promoters, It literally means they are on the "Board of Directors". This means they are definitely PARTNERS.

 

So please, when I have faith in an API aka Vulkan. It is NOT blind faith. Especially when it is created, backed, promoted, etc by the Khronos Group. i have more faith in the Vulkan and OpenGL APIs than in DirectX as the Khronos Group makes Industry Standard APIs. Khronos Group has experience on their side in all aspects of API development. So forgive me if I seem to have "blind faith" in Vulkan. It would seem more like you have "blind faith" in Direct X 12 and Windows 10.