Zink OpenGL-On-Vulkan Looking Quite Good & Shining With Mesa 22.1

[phoronix]

Phoronix: Zink OpenGL-On-Vulkan Looking Quite Good & Shining With Mesa 22.1

Recently I carried out some tests looking at the performance of Zink for OpenGL implemented atop the Vulkan API in a generic manner that works across drivers. With the state of Mesa 22.1, all of Zink's recent improvements are paying off and here is a quick look at where the performance stands against using the RadeonSI OpenGL driver.

https://www.phoronix.com/vr.php?view=31102

[uid313]

Is there room for improvement for Zink?

[microcode]

Is there room for improvement for Zink?

Yes, there has been and there will continue to be room for improvement.

[microcode]

Cool to see Zink winning at high resolution, shows that the shader compilation is working well; and that the rest is close enough that it can lead to a net win in some cases. :+ )

[uid313]

Cool to see Zink winning at high resolution, shows that the shader compilation is working well; and that the rest is close enough that it can lead to a net win in some cases. :+ )

Why is Zink losing at lower resolutions, and can anything be done about it?
If so, what can be done about it, and can you do it?

[STiAT]

Will be interesting if it can squeeze out more performance. Would be interesting if Zink would one day replace all the different OpenGL drivers.

For Mesa users anyway, but in a few years I hopefully can get back to using a card where I can use Mesa ... The nvidia stuff is good, but I am somehow bothered by the drivers not being open.

[Vistaus]

Is there room for improvement for Zink?

There's always room for improvement in any kind of software.

[uid313]

There's always room for improvement in any kind of software.

Well, if we're talking about performance improvements, it's not always much that can be done. Sometimes there is low hanging fruit and much can be done through adding some sort of cache, or analyzing some hot paths, but sometimes there is not much that can be done, maybe it is already heavily optimized.

[darkbasic]

Unfortunately there are still some open source games where Zink is 100x slower than radeonsi. Hopefully one day it will catch up.

[sarmad]

Why is Zink losing at lower resolutions, and can anything be done about it?
If so, what can be done about it, and can you do it?

My guess is that at lower resolutions more frames are rendered and so the Zink overhead accumulates, whereas at higher resolutions the bottleneck is more on the GPU side.