Vulkan seems to be significantly better in multiple ways.
Combine that with recent doubts from certain developers regarding practical limitations of DX12 and it makes Vulkan appear even more advantageous.
I would be very excited as a developer or as a consumer, especially considering the weight Valve is putting behind it.
Maybe eventually one of the very talented CDT members like @computerquip would able to implement this in NS2, seeing as how backwards compatible it claims to be.
Who knows yet how difficult it actually is, though.. X_X
AsranielJoin Date: 2002-06-03Member: 724Members, Playtest Lead, Forum Moderators, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow, Subnautica Playtester, Retired Community Developer
Also keep in mind, Vulcan (and dx12) will be most advantegous if they are used the way they are supposed to be used, which means, multiple threads feeding the GPU with work. This is not the way the ns2 engine is currently designed, simply because dx9 (the main renderer) works best with a single thread. But its indeed exciting. Just needs somebody to actually implement it..
No I think it would be essentialy the same thing, but the discussion is moot because both mantle and vulkan hasn't been released yet. Vulkan will get wider adoption from Khronos though, than Mantle would ever have. That's what is important, and that they will be developing it over time. But that's a political change. Well yes on technical level, for now they swapped HLSL in mantle for GLSL in Vulkan(who cares..) and added multiplatform Intermediate language(more of a bonus for driver developers than added value for end user. In the end Khronos itself would now have to provide GLSL compiler instead and maybe they will enlist help from AMD NV INTEL for that too again, so little change.).
I am not talking about DX12. They will probably try to walk the fine line between optimalizations and backward compatibility.
Good for you. Just implement it, if you feel like it. But you already do DX9, DX11, OpenGL (and PhysX). That's one serious multitasking. Seriously, pick one and stick with it :P Any chance of DX11 comming out of the eternal beta? Maybe do OpenCL, DirectCompute, DX10, Bullet,Havok and CUDA, to have the full set
I think Vulkan will be simple/easy to learn and hard to master. Something like a OpenGL 1/2 -> OpenGL 3/4 switch. They suddenly stoped doing math for you and you did have to learn it. In Vulkan you probably would have to learn how to organize memory and command stream yourself to see the promised benefits.
Yes, there needs to be at least something to save AMD's face.
Either continue Mantle or abandon it and join another project while 'bringing the best ideas'.
Dota 2 on Source 2 engine, running on a toaster with Intel graphics:
^^^ Also: "Would not be possible without Mantle. Thanks AMD" at GDC from Khronos group. Lot of face-saving must be going on... I think others are saving face by accepting it silently under new name.
Either continue Mantle or abandon it and join another project while 'bringing the best ideas'.
Join another project? You are aware how many companies/acedemics and individual contributors are involved with the Khronos group aren't you? It is not something AMD have just joined haha, OpenGL has been around for quite a while now...
If you are not aware who is involved, maybe take a look at this list? I've highlighted a few of the companies you may have heard of.
Khronos current membership Academic
China Academy of Telecommunication Research of MIIT
Columbia University
ETRI
Imperial College London
ITRI
Korea University
National Taiwan University
National Tsing Hua University
Nihon University
Oregon State University
Politecnico di Milano
Seoul National University
Universita di Bologna
University of Bristol
University of Cambridge
University of Toronto Contributor
3D Incorporated Adobe
Advanced Driver Information Technology Altera Corporation Amazon Corporate LLC
Axell Corporation
Axis Communications AB
Beijing Pinecone Electronics Co., Ltd. Blizzard Entertainment Inc Broadcom Corporation
Cadence Design Systems, Inc.
CEVA
Codeplay Software Limited
CogniVue Corporation
Continental Automotive GmbH
Core Avionics & Industrial
Dassault Systemes
Digital Media Professionals Electronic Arts, Inc. Ericsson AB
Freescale Semiconductor Fujitsu Limited
Futuremark Oy
FXGear, Inc. Google, Inc.
Harmonic, Inc.
Huawei Technologies Co. Ltd.
Kaleido Technology
KDAB Ltd.
Kishonti Kft.
Kyungpook National University LG Electronics
Linaro Limited
Los Alamos National Laboratory Lucasfilm Ltd, LLC Matrox Graphics, Inc
MediaTek Inc
Mentor Graphics UK Ltd. Microsoft Corporation
Mobica Ltd
Movidius Ltd Mozilla Corporation
MulticoreWare Inc.
NEC Solution Innovators, Ltd. Oculus VR, Inc.
Oxide Games Panasonic Pixar
Realtime Technology AG
Renesas Electronics
RICOH COMPANY LTD
Rightware Oy
Smith Micro Software, Inc.
Symbio
Synopsys, Inc.
Takumi Corporation Texas Instruments
ThinCI
Think Silicon
Tobii Technology AB Toshiba
TransGaming Inc. Valve Corporation
Verisilicon Holdings, Co. Ltd. VIA Alliance Semiconductor Co., Ltd.
Visteon Corporation VMware, Inc
Wargaming (Austin), Inc.
Xilinx, Inc.
zSpace, Inc.
Executives
Gold Standard Group Individual Contributor
AJ Guillon
Andrew Brownsword
Ben Gaster
Bruce Merry
Dan Ginsburg
Erik Noreke
Fabien Cellier
Fabrice Robinet
Frank Brill
John Kessenich
Jon Leech
Kari Pulli
Mark Callow
Mikael Bourges-Sevenier
Miller & Mattson. LLC
Patrick Cozzi
Perey Research & Consulting
Rob Manson
Tim Lewis
Tomasz Bednarz
Tony Parisi Promoter AMD Apple, Inc. ARM Limited Epic Games, Inc.
Imagination Technologies Intel
NOKIA OYJ NVIDIA Corporation
QUALCOMM Samsung Electronics Sony Corporation
Vivante Corporation
Well, that you are written on some list means little. Microsoft hasn't done much I think. Apple can't even be bothered to update OpenGL to recent version. But AMD, NVIDIA, Intel and some toolmakers and game makers are aboard and actively participating. We can only hope Vulkan will be the one API rules them all again
Oh and ARM seems to be active, so hopefully we will have proper Vulkan in phones and tablets, not some Vulkan ES nonsense.
@Soul_Rider, wow ur jumpy!
Do you know the difference between words 'project' and 'group'?
(btw, I've been here for a long time and coded in Delphi with OpenGL v1)
^^^ Yesss! Do I get argument victory forum badge? >-
Nope!
Because Vulkan is not masquerading as Mantle in disguse, nor is Mantle being ported into OpenGL Next awkwardly - there are some important differences imo.
The game developer can write their source code in pretty much any language they want. Not only does this decrease shader compilation time, because SPIR-V is pre-compiled, but this probably means that a SPIR-V shader will be much more simple than either GLSL or HLSL. It can also disable most of the error checking for production code, rather than having the API look over the game code's shoulder.
Also, unlike Mantle which was only available under AMD catalyst drivers and Microsoft Windows 7 or 8, Vulkan is compatible with any hardware that supports OpenGL ES 3.1, (meaning even smartphones, computer tablets, video game consoles and PDAs - and any desktop GPU from Geforce 400 series onwards) and can support any OS including Linux, iOS, SteamOS, Android - should the graphics vendor choose to support it.
I think Mantle just helped bring the topic to the foreground and spur what was clearly the needed next step in the industry to resolve the untapped potential of our hardware / the current bottlenecks.
Mantle was supposed to be open and inclusive too... They simply gave it to Khronos (and probably Microsoft too), they done few months job on it and announced it as Vulkan, if I put it overly bluntly. They wanted to do new GL when Mantle appeared as anybody else in the industry, but they had nothing at the time. Yes, there are some (necessary) changes HLSL->GLSL. SPIR-V instead of driver dependent IR. And they will continue to change it to make everybody on the board happy. I guess we will see the differences soon enough. Mantle reference should be released this month and Vulkan "later this year"(though we got some sneak peaks here and there what they got now).
Shader compilation is uninportatnt for end user. You need to do it only on installation or GPU/driver change. Ok, we are lazy, let's do it on loading screen (remember that 5 min long loading screen where you load all the textures from disk? You can compile your shaders during that in parallel if you are bothered by the added 2 secs or so). Hell, you can compile it while user is staring at the server list to choose where they gonna play. But hey, it's nice they fixed it this way. I always considered the runtime compiler in GPU driver in OpenGL stupid...
The real bonus is that you don't have to ship your shader source codes. The danger on the other hand is the optimization is spread between the interpreter and compiler. If your game vendor won't update his game, you may not benefit from new GPU and new GPU drivers as much.
PS: They are talking a lot about "compute", so it would be nice if this API would replace OpenCL too in time.
Yes, that's settled. We want Vulkan!
Now what about that quantum processors support and supplanting real player base with true AIs?
On a more serious note, what about those VR helmets support? That looks relatively easy...
BTW, for those sickos who like that kind of stuff (450 pages - admit it you like that, don't ya?), Mantle documentation got public : amd.com/en-us/innovations/software-technologies/technologies-gaming/mantle#downloads
Get a whiff what it and Vulkan will be actually about! As promised it shall deliver more responsibility for the developer and more performance and bugs for the end-user. :P
BTW, for those sickos who like that kind of stuff (450 pages - admit it you like that, don't ya?), Mantle documentation got public : amd.com/en-us/innovations/software-technologies/technologies-gaming/mantle#downloads
Get a whiff what it and Vulkan will be actually about! As promised it shall deliver more responsibility for the developer and more performance and bugs for the end-user. :P
@KrOoze do you mean more performance and less bugs for the end-user, or really more of both??
Both! Certainly untill the drivers and engines using it settle... Also the API as I said gives more responsibility to the dev, that is more opportunities to f up bad, without even knowing you f'ed up bad. On the other hand there seems to be some debugging tools effort. I was always not very happy with debugging tools choices in ol' OpenGL, so maybe they will do better this time...
Bit of a rez here, but I'm gaming on Linux, and NS2 runs good enough for now. I'd love to see Vulkan though so I can get some seriously mad FPS! My Windows brethren meanwhile are having fun getting like 120-200 FPS
Bit of a rez here, but I'm gaming on Linux, and NS2 runs good enough for now. I'd love to see Vulkan though so I can get some seriously mad FPS! My Windows brethren meanwhile are having fun getting like 120-200 FPS
If we add Vulkan support, your "Windows brethren" would also see that speed boost.
Comments
It was just released something. I really did not think it would come.
http://www.techpowerup.com/210328/amd-releases-mantle-programming-guide-and-reference-api.html
No, Mantle is completely different thing.
(and completely useless)
AMD's pretty much encouraging devs to focus on dx12
http://community.amd.com/community/amd-blogs/amd-gaming/blog/2015/03/02/on-apis-and-the-future-of-mantle
AMD(community.amd.com/community/amd-blogs/amd-gaming/blog/2015/03/03/one-of-mantles-futures-vulkan): "The cross-vendor Khronos Group has chosen the best and brightest parts of Mantle to serve as the foundation for “Vulkan,” the exciting next version of the storied OpenGL API."
Vulkan seems to be significantly better in multiple ways.
Combine that with recent doubts from certain developers regarding practical limitations of DX12 and it makes Vulkan appear even more advantageous.
I would be very excited as a developer or as a consumer, especially considering the weight Valve is putting behind it.
Maybe eventually one of the very talented CDT members like @computerquip would able to implement this in NS2, seeing as how backwards compatible it claims to be.
Who knows yet how difficult it actually is, though.. X_X
I am not talking about DX12. They will probably try to walk the fine line between optimalizations and backward compatibility.
Good for you. Just implement it, if you feel like it. But you already do DX9, DX11, OpenGL (and PhysX). That's one serious multitasking. Seriously, pick one and stick with it :P Any chance of DX11 comming out of the eternal beta? Maybe do OpenCL, DirectCompute, DX10, Bullet,Havok and CUDA, to have the full set
I think Vulkan will be simple/easy to learn and hard to master. Something like a OpenGL 1/2 -> OpenGL 3/4 switch. They suddenly stoped doing math for you and you did have to learn it. In Vulkan you probably would have to learn how to organize memory and command stream yourself to see the promised benefits.
Yes, there needs to be at least something to save AMD's face.
Either continue Mantle or abandon it and join another project while 'bringing the best ideas'.
Dota 2 on Source 2 engine, running on a toaster with Intel graphics:
Join another project? You are aware how many companies/acedemics and individual contributors are involved with the Khronos group aren't you? It is not something AMD have just joined haha, OpenGL has been around for quite a while now...
If you are not aware who is involved, maybe take a look at this list? I've highlighted a few of the companies you may have heard of.
Academic
China Academy of Telecommunication Research of MIIT
Columbia University
ETRI
Imperial College London
ITRI
Korea University
National Taiwan University
National Tsing Hua University
Nihon University
Oregon State University
Politecnico di Milano
Seoul National University
Universita di Bologna
University of Bristol
University of Cambridge
University of Toronto
Contributor
3D Incorporated
Adobe
Advanced Driver Information Technology
Altera Corporation
Amazon Corporate LLC
Axell Corporation
Axis Communications AB
Beijing Pinecone Electronics Co., Ltd.
Blizzard Entertainment Inc
Broadcom Corporation
Cadence Design Systems, Inc.
CEVA
Codeplay Software Limited
CogniVue Corporation
Continental Automotive GmbH
Core Avionics & Industrial
Dassault Systemes
Digital Media Professionals
Electronic Arts, Inc.
Ericsson AB
Freescale Semiconductor
Fujitsu Limited
Futuremark Oy
FXGear, Inc.
Google, Inc.
Harmonic, Inc.
Huawei Technologies Co. Ltd.
Kaleido Technology
KDAB Ltd.
Kishonti Kft.
Kyungpook National University
LG Electronics
Linaro Limited
Los Alamos National Laboratory
Lucasfilm Ltd, LLC
Matrox Graphics, Inc
MediaTek Inc
Mentor Graphics UK Ltd.
Microsoft Corporation
Mobica Ltd
Movidius Ltd
Mozilla Corporation
MulticoreWare Inc.
NEC Solution Innovators, Ltd.
Oculus VR, Inc.
Oxide Games
Panasonic
Pixar
Realtime Technology AG
Renesas Electronics
RICOH COMPANY LTD
Rightware Oy
Smith Micro Software, Inc.
Symbio
Synopsys, Inc.
Takumi Corporation
Texas Instruments
ThinCI
Think Silicon
Tobii Technology AB
Toshiba
TransGaming Inc.
Valve Corporation
Verisilicon Holdings, Co. Ltd.
VIA Alliance Semiconductor Co., Ltd.
Visteon Corporation
VMware, Inc
Wargaming (Austin), Inc.
Xilinx, Inc.
zSpace, Inc.
Executives
Gold Standard Group
Individual Contributor
AJ Guillon
Andrew Brownsword
Ben Gaster
Bruce Merry
Dan Ginsburg
Erik Noreke
Fabien Cellier
Fabrice Robinet
Frank Brill
John Kessenich
Jon Leech
Kari Pulli
Mark Callow
Mikael Bourges-Sevenier
Miller & Mattson. LLC
Patrick Cozzi
Perey Research & Consulting
Rob Manson
Tim Lewis
Tomasz Bednarz
Tony Parisi
Promoter
AMD
Apple, Inc.
ARM Limited
Epic Games, Inc.
Imagination Technologies
Intel
NOKIA OYJ
NVIDIA Corporation
QUALCOMM
Samsung Electronics
Sony Corporation
Vivante Corporation
Oh and ARM seems to be active, so hopefully we will have proper Vulkan in phones and tablets, not some Vulkan ES nonsense.
Do you know the difference between words 'project' and 'group'?
(btw, I've been here for a long time and coded in Delphi with OpenGL v1)
Yes, the point is that nobody explicitly states any kind of vendor lock-in.
Because Vulkan is not masquerading as Mantle in disguse, nor is Mantle being ported into OpenGL Next awkwardly - there are some important differences imo.
The game developer can write their source code in pretty much any language they want. Not only does this decrease shader compilation time, because SPIR-V is pre-compiled, but this probably means that a SPIR-V shader will be much more simple than either GLSL or HLSL. It can also disable most of the error checking for production code, rather than having the API look over the game code's shoulder.
Also, unlike Mantle which was only available under AMD catalyst drivers and Microsoft Windows 7 or 8, Vulkan is compatible with any hardware that supports OpenGL ES 3.1, (meaning even smartphones, computer tablets, video game consoles and PDAs - and any desktop GPU from Geforce 400 series onwards) and can support any OS including Linux, iOS, SteamOS, Android - should the graphics vendor choose to support it.
I think Mantle just helped bring the topic to the foreground and spur what was clearly the needed next step in the industry to resolve the untapped potential of our hardware / the current bottlenecks.
Shader compilation is uninportatnt for end user. You need to do it only on installation or GPU/driver change. Ok, we are lazy, let's do it on loading screen (remember that 5 min long loading screen where you load all the textures from disk? You can compile your shaders during that in parallel if you are bothered by the added 2 secs or so). Hell, you can compile it while user is staring at the server list to choose where they gonna play. But hey, it's nice they fixed it this way. I always considered the runtime compiler in GPU driver in OpenGL stupid...
The real bonus is that you don't have to ship your shader source codes. The danger on the other hand is the optimization is spread between the interpreter and compiler. If your game vendor won't update his game, you may not benefit from new GPU and new GPU drivers as much.
PS: They are talking a lot about "compute", so it would be nice if this API would replace OpenCL too in time.
But yea I agree with ya
Remember how NS2 received proper DX11 support in the first place?
Remember how Mantle was added to NS2?
Now what about that quantum processors support and supplanting real player base with true AIs?
On a more serious note, what about those VR helmets support? That looks relatively easy...
Yes absolutely. Only one crash (since DX11 was available) and I'm not sure of it's origin.
Get a whiff what it and Vulkan will be actually about! As promised it shall deliver more responsibility for the developer and more performance and bugs for the end-user. :P
@KrOoze do you mean more performance and less bugs for the end-user, or really more of both??
If we add Vulkan support, your "Windows brethren" would also see that speed boost.