Despite my disclaimer about GC (and my effort to use JMH properly), I find it difficult to trust microbenchmarks on the JVM. I don't know when I'll have time fo

Despite my disclaimer about GC (and my effort to use JMH properly), I find it difficult to trust microbenchmarks on the JVM. I don't know when I'll have time for this, but "someday" I'd like to port this whole codebase to Rust/Criterion (which should be straightforward because the algorithms/data structures are "trivial"), and see if the more surprising artifacts persist. I do find the overall differentiation between RH and BLP surprising; I expected them to have pretty similar performance profiles. In any case, I would definitely appreciate someone else rerunning the benchmarks on a different platform/JVM!