This is a separate crate that contains the criterion.rs benchmarks for Randy. Separated here because it increases the f* out of compile times, and dev-deps arent optional

For repeatability you should consider setting your frequency scaling gov to performance and pinning the run to a particular CPU with taskset. example:

sudo cpupower frequency-set -g performance

cargo bench --norun     # compile with all CPUS
taskset 0x4 cargo bench # pin bench to CPU 0x4

I wrote a wrapper macro that I try to almost fully disable at compile time... unless the timings feature is on. I want to see if the pass through impl of the macro causes any noticeable perf diff. Seems like NO :D example:

cargo bench --bench macro

Benchmarks for various proc_info (top like) methods.

To run set TPID var to something like /proc/1234. example:

TPID="/proc/`pgrep emacs | head -n1`" cargo bench --bench proc