Mesa Git Is Now Officially Mesa 11.0

[smitty3268]

As for Mesa ganing on OpenGL spec take a look at this

OpenGL Mesa-support
1.3 2001 = same year
1.4 2002 = same year
1.5 2004 = 1 year late
2.0 2007 = 3 year late
2.1 2007 = 1 year late
3.0 2012 = 4 year late
3.1 2012 = 3 year late
3.2 2013 = 4 year late
3.3 2013 = 3 year late
4.0 2015 = 5 year late
4.1 2015 = 5 year late

Doesn't exactly look like they are gaining on the spec does it :/


I would say 4.5 would be here 2016/2017 at the earliest, the good news there is that it would put Mesa back to being only 3 years behind the spec.

Indeed, though that's somewhat misleading to lean on those numbers too strongly. In fact, the GL spec itself was lagging far behind what the hardware was capable of.

For example, when 3.3 was finally supported by Mesa in 2013, not only was it 3 years late of the spec but 7 years behind the hardware capabilities. The binary drivers were able to move the spec forward very quickly at that point, because their drivers already supported the more advanced functionality from the DirectX side. Mesa was truly far behind at that point, and has been catching up ever since.

There's no reason to think OpenGL 4.5 won't be finished up in 2016 at this point given how much of the remaining spec has already been done and posted as in progress patches. Plus, OpenGL 4.3 is all that's really interesting at the moment. 4.4 and 4.5 are already mostly done, with the remaining bits mostly trivial or very, very limited in use.

[Ancurio]

Cool! Some info about what is the plan/roadmap about tesselation and fp64?

WIP r600 tessellation: http://cgit.freedesktop.org/~airlied.../?h=r600g-tess
Note sure about fp64.

[zanny]

Indeed, though that's somewhat misleading to lean on those numbers too strongly. In fact, the GL spec itself was lagging far behind what the hardware was capable of.

For example, when 3.3 was finally supported by Mesa in 2013, not only was it 3 years late of the spec but 7 years behind the hardware capabilities. The binary drivers were able to move the spec forward very quickly at that point, because their drivers already supported the more advanced functionality from the DirectX side. Mesa was truly far behind at that point, and has been catching up ever since.

There's no reason to think OpenGL 4.5 won't be finished up in 2016 at this point given how much of the remaining spec has already been done and posted as in progress patches. Plus, OpenGL 4.3 is all that's really interesting at the moment. 4.4 and 4.5 are already mostly done, with the remaining bits mostly trivial or very, very limited in use.

The last major frontier is compute shaders it seems, and Intel has a vested interest in GLES 3.1, so that will probably happen a lot faster than Tessellation did.

[dungeon]

Indeed, though that's somewhat misleading to lean on those numbers too strongly. In fact, the GL spec itself was lagging far behind what the hardware was capable of.

For example, when 3.3 was finally supported by Mesa in 2013, not only was it 3 years late of the spec but 7 years behind the hardware capabilities.

Not sure what is misleading, how behind hardware capabilities are DX12 or Vulkan... By your definition, they lagging 4 years already

[Anvil]

I can guarantee with certainty that Ubuntu 15.10 will have Mesa 11, along with Linux 4.2. LLVM 3.7 releases this month, which means that yes, it will also be in 15.10.

got a Source ? to prove that

[eydee]

Comparing how long Mesa got version support in the past and present in not a good idea. Linux gaming did not exist until only a few years ago. There was nothing to drive the development, there was no demand at all. There were also no open source drivers that could make use of it and produce playable game performance. Things have changed though.

[przemoli]

It would make more sense to see extensions per month.

Some extensions require a lot of work. Not all games require all though. So there is case for going partial.

[justmy2cents]

what will they do once OGL is feature complete? optimise for performance/ bug smashing? DirectX State Tracker? OpenCL? HSA?

optimizing is sure bet along the way. my best bet would be that they'll be focusing on vulkan renderer for opengl like llvmpipe or softpipe. if they do that, then Mesa will become defacto longterm solution for compatibility on anything supporting Vulkan.

directx is another thing that would be much more suited outside of Mesa and based on Vulkan as well. Vulkan is at same level as galium and at the same time it will be supported by both worlds, oss and proprietary unlike galium nine which only works on galium

[DrYak]

Comparing how long Mesa got version support in the past and present in not a good idea. Linux gaming did not exist until only a few years ago. There was nothing to drive the development, there was no demand at all. There were also no open source drivers that could make use of it and produce playable game performance.

...in the recent past.
In the distant past, Mesa was one of the few solution to get OpenGL 1.x running on hardware that has been more or less abandonned by the parent company.

There were people who managed to get Doom3 running on 3DFx Voodoo (recent like Voodoo5 but even older like Voodoo2) cards. That's a OpenGL 1.4 game running on hardware that isn't supposed to support beyond OpenGL 1.1
By that time 3DFx had folded and got bought up by Nvidia. You couldn't hope to get much openGL support from them. But you still had batshit crazy devs hacking MesaFX until Doom3 was able to run. (Although with ugly visual quality, due to missing shader capabilities, but still work).

Might, too, explain why back then there wasn't that much lagging behind as Mesa got support for versions of 1.x
There was actual demand for it and enough enthusiasts.