MiMalloc 3 Comparison

[SchrodingerZhu]

https://github.com/microsoft/mimalloc

Seems v3 brings in some new stuffs such as numa affinity and performance improvement.

[mjp41]

I think many of the things that have been improved in mimalloc 3 are hard to benchmark. Large long running systems, so a empirical comparison with mimalloc-bench wouldn't say very much.

[SchrodingerZhu]

One thing that may be interesting to think about is to benchmark against lean4's mathlib. That is a huge project with a lot of RC flying around given the feature of Lean4's runtime.

[mjp41]

Could we add that to mimalloc-bench? How long does it run for?

[SchrodingerZhu]

It runs for around 1 hour without elaboration cache.