Working Towards Lower OpenGL Driver Overhead

[Ancurio]

DX12 is vaporware. So let it arrive first.

http://blogs.msdn.com/b/directx/archive/2014/03/20/directx-12.aspx

[przemoli]

DX12 should likely eliminate any performance advantages OGL currently has.

That depend on MS willingness to force GPU makers into one vision.

Right now?

They (MS) lag behind OpenGL advancemnts.
They (MS) seam to be weaker at pushing for unified feature set. (ARB release more ARB extensions, then MS its own "extensions")

And MS was occupied with Xbox One.


My guess is that original MS plan was to backport changes to DX to Win (minus those that do not make sense).
But low level access like mantle? (You compose Your own command queues and assign them to the GPU parts, You do it on any core, etc.) No way. It will be more like OpenGL. Where You just avoid changing state too often.

[przemoli]

http://blogs.msdn.com/b/directx/arch...irectx-12.aspx

GREAT LINK!

DX12 talk at GDC will be PR. But that blog at least clearly state that DX12 will be real thing...

and will be vapoware for 2y

"""
Q: When will I be able to get my hands on DirectX 12?
A: We are targeting Holiday 2015 games.
"""

OpenGL rulez. (Or will, if MS will indeed take its sweet time to release DX12. 2y is more then enough for OpenGL to regain most of the ecosystem even on Windows)

[hajj_3]

That depend on MS willingness to force GPU makers into one vision.

Intel and Nvidia have pledged support for DX12. AMD hasn't made any announcements.

[smitty3268]

Intel and Nvidia have pledged support for DX12. AMD hasn't made any announcements.

No doubt they're trying to formulate their response to the inevitable "what does this mean for mantle" question that will come up when they announce their support. It's obvious that DX12 is directly aimed at replacing the need for Mantle.

But AMD will obviously support it. They've always been good at supporting Direct3D versions, and not having that checkmark for their marketing team wouldn't fly. Especially if games start coming out using it and really are faster - starting to lose benchmarks would be a disaster.

[vein]

there is no point in mantle on linux. right now, their opengl is still subpar and i would prefer them fixing that 1st and then add another unfinished layer. adding subpar mantle which would be probably beta until 2019 would just add to bugs on amd unless you prefer having buggy gl+beta mantle

Well...considering the fact that 2 major game engines are going to give full support on mantle (Frostbite and Cryengine), it would be a dream to get mantle on linux. And appearantly AMD is working on it... one can only hope that they will make it good and not a bad beta version...

[tomtomme]

DX12 should likely eliminate any performance advantages OGL currently has.

DX12 is scheduled for the end of 2015. Almost 2 years are a long time for OGL (or others) to add new advantages...

[AnonymousCoward]

well it seems like DX12 is ripoff Mantle https://twitter.com/repi/status/446787503953944576

[curaga]

Gotta share the love:

http://deslide.clusterfake.net/?o=html_table&u=http%3A%2F%2Fwww.slideshare.net%2FCassEveritt%2Fapproaching-zero-driver-overhead%23

If this is the first time you hear of deslide, share the good to at least one other place where slideshare is mentioned.

[erendorn]

Well...considering the fact that 2 major game engines are going to give full support on mantle (Frostbite and Cryengine), it would be a dream to get mantle on linux. And appearantly AMD is working on it... one can only hope that they will make it good and not a bad beta version...

Do you really think a game would be released for a platform where only one card manufacturer would be supported? AMD porting mantle to Linux is of zero use with respect to having Frosbite games on Linux as long as Frosbite does not support some other API available for Intel and Nvidia GPU on Linux (that, or Intel and Nvidia support of mantle on Linux, which might not happen just tomorrow).