Question Speed up NativeList allocation/initiation

I am working on a system where I have to make arrays of 60k items, one for vertex position, normal, and UV.
This takes up +- 2ms, which is too much for my current setup. Is there a way to improve this performance?
Or is there maybe a way to get data from a mesh into a combined array (float3x2 for example for position and normal combined)?

 

GetVertexData should do what you want, but its going to depend on how mesh is structured.
Also, do you allocate lists every frame? Is there a way to not do that maybe? Do you set initial capacity to reduce amount of reallocations of the list?

 

You only need one array normally because you can make a struct containing position, normal, UV and whatever you need. See link by Saniell and move up one level to Mesh.MeshData and explore that a bit.

You could also alloc that array with persistent allocator and keep it around if you can afford to block this much memory (probably 1-2 megabytes).

 

Thanks! This is helping a lot with the performance.
Looking more into the MeshData API as well to improve performance even more (but it's a headache since the documentation is a bit limited sometimes)