![]() Khronos’s last attempt to create a new, non-backwards-compatible spec - Longs Peak - failed dismally, so it’ll be keen to not make the same mistakes twice. The Khronos Group, which develops the OpenGL spec (and other related specs), says it’s working hard with all of its consortium members to develop a spec that hardware and software companies will actually use. This didn’t matter so much in the olden days, when GPUs were nasty things that needed high-level abstraction to be programmed sensibly - but now, as GPUs become ever more mature and sane and documented, developers are asking for graphics APIs that allow them to get much closer to the bare metal, significantly improving performance and reducing overhead.Įnter OpenGL NG, which will be a ground-up redesign of OpenGL that isn’t backwards compatible. More importantly, though, OpenGL - again, like DirectX and Direct3D - is a very high-level API that makes it hard to efficiently run code on the GPU directly. Much like DirectX (which is only a little younger), OpenGL has gained a lot of bulk and scar tissue over the years. With more than 22 years under its belt, OpenGL (originally released by SGI in 1992) is the oldest high-level 3D graphics API still in popular use. Khronos has an uphill struggle ahead, though: While AMD and Microsoft are focusing on their own specific implementations, OpenGL NG will be a cross-platform solution for all operating systems and hardware makers, just like the existing OpenGL specs. ![]() The idea, much like AMD’s Mantle and DirectX 12, is to build an entirely new version of OpenGL that removes a lot of the abstraction, significantly reducing the overhead and inefficiencies when working at a low level with the bare metal GPU hardware. Next Generation OpenGL (OpenGL NG), however, is a complete rebuild of the OpenGL API. OpenGL 4.5, except for some new Direct3D 11 emulation features for easier porting, is your fairly standard annual OpenGL update. OpenGL 4.4 being a small update, NVIDIA had no trouble to ship a good OpenGL 4.4 implementation at Siggraph.At Siggraph 2014, the Khronos Group has announced both OpenGL 4.5 and, more excitingly, the Next Generation OpenGL Initiative. Mesa is a bit of a disappointment considering that geometry shader is still not supported so we are bound to OpenGL 3.1 on this platform. I wouldn't be surprised if Intel manages to release at Siggraph 2014 an OpenGL implementation on part with latest specifications at the time. According to my sample tests, the implementation works pretty well and it also provides the big OpenGL 4.3 extensions:Ĭonsidering how far Intel was just a couple of years ago, it's fantastic. Intel continues to work hard on its OpenGL implementation reaching OpenGL 4.2. ![]() It seems that AMD has released an OpenGL 4.3 driver but as I don't have an AMD card at the moment so I haven't tested it.Īlso, according to Brano Kemen, Catalyst 13.8 causes a performance drop and glClientWaitSync is particularly slow (> 7ms in his case). However, considering that Apple took 3 years for that.Īt least, it's great for OpenGL programmers because it means that we will be able to target OpenGL 4 hardware. Going from OpenGL 3.2 to OpenGL 4.1 for Apple is great.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |