whiterabbit42k

Re density map, interesting! Yea so I have actually switched to `sequential_generate_sparse_density_map` for now, the perf difference is on average quite large (for me, i do have a lot of...

> I think dropping the maps by using a different marching cubes strategy would help a lot as you could really need the cache efficiency here. Sounds like a plan...

sorry for the delay! so just rebased, and seeing these numbers for about 500 particles: ``` Reconstructed mesh size: vertices=8352 indices=8352 reconstruct_surface_inplace: 100.00%, 14.67ms/call @ 68.18Hz compute minimum enclosing aabb:...

So I noticed the inline(never) annotation everywhere, as mentioned in the inout branch. Is the rationale there for the profiling infra to have more accurate reporting? If yes, perhaps we...

let me try with update, thanks for asking