Next-Gen OpenGL To Be Announced Next Month

[profoundWHALE]

I don't know that Mantle and Metal are any lower-level than, say, using OpenGL extensions for a particular piece of hardware.

Extensions are extensions. I was going to add it on the list like this:

Low Level: Mantle, Metal, OpenGL Extensions, Directx 12
Mid Level: OpenGL, Directx
High Level: SDL? Irrlicht?

Basically it's not fair to add Dx12 since the next OpenGL hasn't been announced and OpenGL extensions are extensions, meaning we still have the whole mess of which way to do something and old legacy code.

[CrystalGamma]

btw arm and x86 gaming market has very few overlaps, they dont use opengl they use openal as primary api, that maybe could be a way between opengl garbage and mantle a middle way.

OpenAL is a SOUND API("Open" Audio Library). Get your facts straight.

[log0]

This already exists. It's called SDL / SFML / Allegro / L?ve2D.

SDL is not an abstraction for 3D rendering. It has some 2D functionality, but that's it.
SFML is pretty much SDL in C++.
Allegro seems to have some OpenGL/Directx wrapper. I don't think it abstracts much though (my first impression).
L?ve2D is 2D only? LUA based game engine...

There are FOSS 3D engines out there like Ogre3D, Urho3D, Panda3D etc. But they all have their pros and cons, varying features, performance and complexity.

[johnc]

I don't know that Mantle and Metal are any lower-level than, say, using OpenGL extensions for a particular piece of hardware.

[Ancurio]

The way people are talking here it sounds like there wasn't. My mistake.

So I'm assuming it goes like this:
Low Level: Mantle, Metal
Mid Level: OpenGL/Directx
High Level: SDL

Almost. Mantle, Metal, OpenGL and D3D are all on the same tier.

[profoundWHALE]

This already exists. It's called SDL / SFML / Allegro / L?ve2D.

The way people are talking here it sounds like there wasn't. My mistake.

So I'm assuming it goes like this:
Low Level: Mantle, Metal
Mid Level: OpenGL/Directx
High Level: SDL

[Ancurio]

Would a 2 tier system work? By that I mean have a very high level, fast to code for and get going version. Then the next tier would be faster but take longer to code for. The bonus is that it should store both tiers at the same time, but you specify which to use in case something goes wrong.

This already exists. It's called SDL / SFML / Allegro / L?ve2D.

[profoundWHALE]

In GL's case, it means you have a selection of N ways to do any one thing. Also in GL's case, the most obvious, perfect fit that is simple to use and only requires a couple lines, is the slow and deprecated one, whereas the performant one is complicated to use and requires hundreds of lines to implement.

Perhaps you can see the problem now No matter what they do, they'll anger devs. Remove the deprecated and slow functionality - you just made programming in GL harder for everyone. Don't remove them - devs are angry because the obvious way is slow.

Would a 2 tier system work? By that I mean have a very high level, fast to code for and get going version. Then the next tier would be faster but take longer to code for. The bonus is that it should store both tiers at the same time, but you specify which to use in case something goes wrong.

[pijaczek]

You mean DX10.1 right? IIRC it's because AMD was the first one to support it by a long shot, there where a few games that implemented it, even some that removed the capability because Nvidia paid them off as DX10.1 made the games run noticeably faster then DX10.

As far as I remember Nvidia paid for use postprocess AA shaders (that was faster with Dx10 then Dx10.1/GL3.2 with MSAA access from shaders) - now every card give access for msaa from shaders (on GL3.2 from GF8800), but nobody cares, and use postprocess AA (and can use it with dx9/gles) like FXAA, MLAA, TXAA... becouse it's faster.

[curaga]

Just out of curiosity, from a developer's perspective, what does the "legacy cruft" matter? Other than running into outdated documentation on the web, when would developers ever encounter it?

In GL's case, it means you have a selection of N ways to do any one thing. Also in GL's case, the most obvious, perfect fit that is simple to use and only requires a couple lines, is the slow and deprecated one, whereas the performant one is complicated to use and requires hundreds of lines to implement.

Perhaps you can see the problem now No matter what they do, they'll anger devs. Remove the deprecated and slow functionality - you just made programming in GL harder for everyone. Don't remove them - devs are angry because the obvious way is slow.