Peer review source code

[watson]

The benchmarks might be showing wrong results and need to be peer reviewed.

Area worth paying special attention to: The V8 JavaScript engine might be optimizing the some functions so the benchmark isn't actually running the expected code.

[jchook]

Hi, author of uuid-random here.

Do you have any more info on how V8 does this? In my experience, "accessing" system entropy seems like the largest performance bottleneck, which hints to why caching gives such a massive speed boost.

I recently noticed my benchmarks quintupled in ops/sec between node v10 and node v12. I would love to know what changed and why that happened.

I did not see a comparable speedup in, e.g., id128.