[libcxx-commits] [libcxx] [libc++] Implement generic associative container benchmarks (PR #123663)

[Louis Dionne via libcxx-commits]

ldionne wrote:

> I am going to run all these benchmarks overnight with `->Repetitions(5)` to see whether they have a lot of variance.

I found that some operations on unordered containers had a lot of variance and I tweaked the benchmarks to reduce the variance to below 15-20% in all cases, which IMO is the best we'll get. Most benchmarks have significantly less variance than that, though (like less than 1%).

https://github.com/llvm/llvm-project/pull/123663