Mesh.SetVertices slower with float3 input

I am working on a bursted mesh slicer, and noticed that mesh.SetVertices is a lot slower if you input an array of float3 (0.57ms vs 0.12ms).
Is there any way this can be improved? This is making it very hard to work with burst for me

Turns out IL2CPP improved performance a lot

The timings mentioned (0.57ms vs 0.12ms) is only spent in mesh.SetVertices and does not capture anything else? With IL2CPP you do not see a difference?

Yes, but with deep profiling enabled.
With IL2CPP it was so small I could not quickly find it, so it was a non issue for this.

If you find this an interesting find I can post more info/make a bug report tomorrow

Always feel free to report a bug. Just to manage expectations; the case might get closed with a simple explanation or deprioritized to fix other bugs that have a higher user pain.

1 Like