Khronos Publishes Its Slides About OpenGL-Next

[johnc]

Arguably, couldn't you say the current state of openGL is a train wreck? I guess a more accurate way to describe it is it's a train that is still on it's rails but has a failing engine.

I personally thing OpenGL-next sounds fantastic, though I'm a little disappointed that it breaks all compatibility. This is going to be a burden for Mesa and open source driver developers, and I'm not sure how this will pan out for people with older hardware. I think it's also potentially an issue for other projects too like Blender, Qt, VLC, and other things that have a pretty deep openGL implementation. I'm not too familiar with how to code for openGL but I can't imagine its a simple as just tweaking a few lines here and there and suddenly everything works. But, perhaps they may offer some backward compatibility, where some of the old obsolete code just doesn't do anything.

That's my concern too, that there is now going to be an "old" OpenGL and a "new" OpenGL, which is going to be a hell of a mess.

Developers won't have to deal with that kind of mess in DX; score another point for DX.

And it's not really obvious what benefits it's actually going to bring, other than chasing meaningless marketing buzzwords like "low-level" or the mythical "everything is better when you start over from scratch".

[MartinN]

- OpenGL-Next is a ground-up design of a modern 3D+Compute API and will break compatibility with existing OpenGL implementations.

- This new initiative is "NOT" going to be a "multi-year, design-by-committee process."

I like the sound of this. These two visionary declarations is what will make this next version of OpenGL successful and competitive, more so than anything else.

[kwahoo]

I am quite surprised Apple is in the list.

Why? Metal is designed for mobile, tile-based rendering GPUs. They still need something different for desktop.

[agd5f]

Developers won't have to deal with that kind of mess in DX; score another point for DX.

I'm not really sure you can say that without knowing what DX12 will entail. This new GL is not really a incremental compatible change like DX10 and DX11 or GL 3.x and 4.x. Making it compatible would basically remove all the advantages it provides over the old interfaces.

[vadix]

I like the sound of this. These two visionary declarations is what will make this next version of OpenGL successful and competitive, more so than anything else.

Yeah, we needed this. However, I also think unifying OpenGL ES and OpenGL was just as important a declaration.

[blackout23]

Arguably, couldn't you say the current state of openGL is a train wreck? I guess a more accurate way to describe it is it's a train that is still on it's rails but has a failing engine.

I personally thing OpenGL-next sounds fantastic, though I'm a little disappointed that it breaks all compatibility. This is going to be a burden for Mesa and open source driver developers, and I'm not sure how this will pan out for people with older hardware. I think it's also potentially an issue for other projects too like Blender, Qt, VLC, and other things that have a pretty deep openGL implementation. I'm not too familiar with how to code for openGL but I can't imagine its a simple as just tweaking a few lines here and there and suddenly everything works. But, perhaps they may offer some backward compatibility, where some of the old obsolete code just doesn't do anything.

It could be that it will only be available on DX11 hardware anyway. Kind of like DirectX 12 (NVIDIA will support it on every DX11 GPU they said) or Mantle (there are reports that DX11 hardware features are required, but may not be enough for it). For NVIDIA this would start with the 400-series (Fermi). OpenGL 4.x also isn't available for old hardware.

Why? Metal is designed for mobile, tile-based rendering GPUs. They still need something different for desktop.

On the desktop they also don't have their own SoC, but use AMD/NVIDIA GPUs. Metal is only for their own SoC.

[Ziple]

This time they have understood that Direct3D is their model. Good point. Just ten years late.

[MartinN]

Yeah, we needed this. However, I also think unifying OpenGL ES and OpenGL was just as important a declaration.

You're correct. I'm not a game developer (or an OpenGL dev in any other way) - and I'm typically non concerned about the technical aspect as that usually gets smoothed out well with competent people, which these are. Breaking with the past, and re-inventing/transforming these APIs, with as wide input and feedback as we can summon, in as agile manner as possible, is what I believe is a key ingredient for continued success... not just one hit wonders that rest on their laurels for 20+ years, patched beyond recognition....

listen...redesign..reinvent...transform.... produce....

[justmy2cents]

Arguably, couldn't you say the current state of openGL is a train wreck? I guess a more accurate way to describe it is it's a train that is still on it's rails but has a failing engine.

I personally thing OpenGL-next sounds fantastic, though I'm a little disappointed that it breaks all compatibility. This is going to be a burden for Mesa and open source driver developers, and I'm not sure how this will pan out for people with older hardware. I think it's also potentially an issue for other projects too like Blender, Qt, VLC, and other things that have a pretty deep openGL implementation. I'm not too familiar with how to code for openGL but I can't imagine its a simple as just tweaking a few lines here and there and suddenly everything works. But, perhaps they may offer some backward compatibility, where some of the old obsolete code just doesn't do anything.

they announced "old" will be supported and updated for quite some time.

/*warning, just my guts speaking*/ although, if new one is low enough, i think it wouldn't be far fetched if we saw singular "old" implementation based on "new" one. it would probably lose some speed, but it could lead to universal implementation that really works as it should. somewhat like mesa software pipe, just working over "new" gl and being accelerated

[clementl]

I like the sound of this. These two visionary declarations is what will make this next version of OpenGL successful and competitive, more so than anything else.

Especially since there are a few gaming focused businesses involved, like Valve, Epic Games and Unity. Those guys will certainly strive for a easy-to-use API.