1 unstable release

0.3.0 Jan 27, 2023

#340 in Profiling

MIT license

320 lines

Run function repetitively and prints out aggregated result.

Differences from other benchmark library:

  • Do not enforce measuring "wall time". It could be changed to "CPU time", "memory usage", etc.
  • Do not run only the benchmark part repetitively. For example, a benchmark needs some complex setup that cannot be reused across runs. That setup cost needs to be excluded from benchmark result cleanly.
  • Minimalism. Without fancy features.

No runtime deps