How to get legitimate micro-benchmark timings in Zig

[muted leaf]

How to get legitimate micro-benchmark timings in Zig

The modification of the sample was an attempt to make it so the optimizer couldn't think the calculation was redundant

[mortal marlin]

Might help to see how you're using std.time.Timer.

Using https://github.com/karlseguin/zul/blob/master/src/benchmark.zig which is just a wrapper around std.time.Timer, I get:

crc24
  9870087 iterations    298.61ns per iterations
  0 bytes per iteration
  worst: 459ns    median: 292ns    stddev: 21.28ns

crc24Openpgp
  27482973 iterations    108.92ns per iterations
  0 bytes per iteration
  worst: 250ns    median: 125ns    stddev: 23.41ns

noop
  101440083 iterations    29.15ns per iterations
  0 bytes per iteration
  worst: 209ns    median: 41ns    stddev: 20.33ns