Are OpenGL and Vulkan good to use at such low level operations?

Haswellx86 · February 8

It was hard for me to find at the basic level, which component in Windows is responsible to just running the display (AI and the MS documentation is confusing me with GDI, Direct2D, Direct3D, etc), but Windows has its own suite for graphics rendering called DirectX, optimized for their own operating system. Apple has Metal on the other hand for their operating system. However for most UNIX based operating systems, there is no proprietary or specifically made for graphics rendering API. We mostly have only OpenGL and Vulkan in the UNIX world.

Even though we have GTK or QT, X11 or Wayland, compositors, other frameworks and all that, they all use OpenGL or Vulkan in the backend and so at the basic level the API which is used to just run your display will be either OpenGL or Vulkan. This are graphics APIs capable of complex 3D graphics. It would be like simulating basic 2D graphics in a 3D environment.

Or do we just see these more like just GPU programming APIs? Because you can use either for GPU compute, not that is probably is going to be the most efficient way than a GPGPU API like OpenCL or CUDA. Either way, I know at least OpenGL is outdated and has more CPU overhead, but Vulkan adoption has been slow.

LloydLynx · February 8

1 hour ago, Haswellx86 said:

Or do we just see these more like just GPU programming APIs?

It's kinda more like that. They work well for both complex 3D rendering and basic desktop acceleration. Though I haven't seen Vulkan used for desktop acceleration due to the vast amount of OpenGL only hardware there is still in use.

riklaunim · February 9

If you want modern and fancy stuff, low level/close to hardware then you go Vulkan, other cases OpenGL. Games or other similar software in the end will use DX12 on Windows and Metal on macOS while Linux will just use Proton/Wine to wrap DX calls and translate them to Vulkan, while on macOS wrappers will translate Vulkan or DX into metal.

And rarely people interact with those APIs directly, they use various engines that implement features on top of them that allow getting the job done quicker.