Conversation

Replying to
That's a microbenchmark of the malloc implementation using various size classes and writing data to the memory allocations. It's a drastic performance improvement since it's a perfect use case for the library. It only needs to set up the two divisors for each size class at init.