Zink Can Now Run On Lavapipe But You Really Want To Avoid It

In addition to this week seeing Zink now running on NVIDIA's proprietary driver for supporting this Gallium3D-based OpenGL over Vulkan implementation, it can now run on top of Lavapipe as the CPU-based Vulkan implementation. But for end-users that is really something you would want to avoid.

With the newest Mesa 21.1-devel code as of today, there is support for running Zink on Lavapipe. Lavapipe is the software Vulkan implementation merged last year to Mesa as a CPU-based Vulkan software driver. So with the latest Mesa Git code, it's possible to route Zink through Lavapipe to ultimately render using the CPU.

This was added so Zink with Lavapipe can be used as part of continuous integration (CI) testing. But beyond that there isn't much real benefit.

For those wanting CPU-based OpenGL, it's better to just use LLVMpipe straight away and avoid the extra indirection and overhead of routing through Vulkan.

Even with today's greatest processors, the likes of Lavapipe and LLVMpipe are still painfully slow and not really relevant for much more than testing along a vendor-neutral code path or the likes of a fallback for a composited desktop. For those thinking they are going to be gaming using LLVMpipe/Lavapipe are dreaming with any mundane working GPU coming out faster in that case than today's best CPUs. In any case, fun to note that Zink can now run on Lavapipe just that you wouldn't want to be doing it outside of testing / CI purposes.