Vulkan/OpenGL-Next the future of OpenGL and Games

[jjmafiae]

at GDC last week they show cased the successor to OpenGL and Mantle, which is an highly optimized low level API.

http://www.performancepsu.com/vulkan-ne ... rs-opengl/

http://www.kitguru.net/components/graph ... nd-drones/

http://www.extremetech.com/gaming/20083 ... -microsoft

now is the question could love in the future get support for this ?

[BluBillz]

If you are asking if love2D will be compatible for OpenGL then that would be pretty neat to see...but i don't see how it could be that much useful...Love2D is powerful for what it is!

[s-ol]

If you are asking if love2D will be compatible for OpenGL then that would be pretty neat to see...but i don't see how it could be that much useful...Love2D is powerful for what it is!

You mean whether it will be ported to Vulkan. Löve is currently running atop the OpenGL graphics stack.

That being said I think it's very possible we get a newer GPU Api, but I'm not sure whether Vulkan will be the new standard, there might be a similar project emerging. OpenGL (and extensions) are a mess right now, and similarly to how X11 is slowly being replaced by Wayland.

[slime]

It's Vulkan, not Vulcan.

now is the question could love in the future get support for this ?

I don't have any plans to do that. Vulkan (and Metal and DirectX 12, which are all very similar) has high system requirements. My development laptop has an OpenGL 4.5-capable GPU but won't support Vulkan (on Windows and Linux at least) since its GPU's architecture is rather old and AMD isn't interested in creating Vulkan drivers for its pre-GCN architectures.

LÖVE wouldn't get huge performance gains from using Vulkan either, compared to OpenGL. Many of the performance improvements possible with the newer APIs (DX12, Vulkan, Metal) can only happen if the code calling those APIs has high-level knowledge of what each frame is going to do. LÖVE's API is intentionally not very controlling, so it doesn't have a whole lot of high-level knowledge compared to game engines.

[Jeeper]

It's Vulkan, not Vulcan.

now is the question could love in the future get support for this ?

I don't have any plans to do that. Vulkan (and Metal and DirectX 12, which are all very similar) has high system requirements. My development laptop has an OpenGL 4.5-capable GPU but won't support Vulkan (on Windows and Linux at least) since its GPU's architecture is rather old and AMD isn't interested in creating Vulkan drivers for its pre-GCN architectures.

LÖVE wouldn't get huge performance gains from using Vulkan either, compared to OpenGL. Many of the performance improvements possible with the newer APIs (DX12, Vulkan, Metal) can only happen if the code calling those APIs has high-level knowledge of what each frame is going to do. LÖVE's API is intentionally not very controlling, so it doesn't have a whole lot of high-level knowledge compared to game engines.

If we got you a new computer, would the answer still be the same?

If not, then I suggest we start a donation campaign to fix that.

[slime]

The rest of what I said still applies (and the high system requirements would be a problem for users of LÖVE and people who want to play LÖVE games as well), so I still wouldn't have plans to do that.

[jjmafiae]

The rest of what I said still applies (and the high system requirements would be a problem for users of LÖVE and people who want to play LÖVE games as well), so I still wouldn't have plans to do that.

can't we have old versions as well as new versions of OpenGL in the engine and use what is possible to use on each computer?

like say a user has a decent Vulkan enabled system then it would automatically switch to that but if the users is still let's say use OpenGL3 then it switches to that ?

[adrix89]

The rest of what I said still applies (and the high system requirements would be a problem for users of LÖVE and people who want to play LÖVE games as well), so I still wouldn't have plans to do that.

can't we have old versions as well as new versions of OpenGL in the engine and use what is possible to use on each computer?

like say a user has a decent Vulkan enabled system then it would automatically switch to that but if the users is still let's say use OpenGL3 then it switches to that ?

There is no point.
It all boils down to performance and game makers probably will not push too much in that direction so that its available and the widest devices.

[slime]

can't we have old versions as well as new versions of OpenGL in the engine and use what is possible to use on each computer?

Yes, and LÖVE already does that to an extent.

Vulkan is not OpenGL though. They're completely separate and independent APIs covering different levels of abstraction of the GPU hardware. The difference between OpenGL and Vulkan is similar to the difference between love.graphics and OpenGL, in many ways.

LÖVE could have a Vulkan backend separate from its OpenGL backend, but maintaining two separate graphics backends would be a lot of extra work even after the Vulkan backend is "complete" (not to mention anyone wanting to contribute to LÖVE's source code would have to be able to program for OpenGL and Vulkan if they want to touch love.graphics.)
And even if that were all done, what would be the gain? a little bit of extra CPU performance on systems that already have high-end specs and aren't bottlenecked by OpenGL's overhead in LÖVE games?

[T-Bone]

If speed is the issue, I think there are more promising changes that could be made, like running on ANGLE by default on Windows (possible with Löve 0.10.0 which supports OpenGL ES natively), which would improve performance on computers with poor OpenGL driver support. Something like this would, unlike Vulkan support, really improve performance on devices that actually need it.

That said, if you feel like making a Vulkan port then go straight ahead! It might not get merged into the main Löve builds, but it could still be a fun project.