View Source erlperf_cli (erlperf v2.2.0)

Command line interface adapter.

Exports functions to format erlperf:benchmark/3 output in the same way as command line interface.

   #!/usr/bin/env escript
    %%! +pc unicode -pa /home/max-au/git/max-au/erlperf/_build/default/lib/erlperf/ebin
   main(_) ->
       Report = erlperf:benchmark([
           #{runner => fun() -> rand:uniform(10) end},
           #{runner => {rand, mwc59, [1]}}
       ], #{report => full}, undefined),
       Out = erlperf_cli:format(Report, #{format => extended, viewport_width => 120}),
Running the script produces following output:
   $ ./bench
   Code                                                ||   Samples       Avg   StdDev    Median      P99  Iteration    Rel
   {rand,mwc59,[1]}                                     1         3  80515 Ki    0.59%  80249 Ki 81067 Ki      12 ns   100%
   #Fun<bench__escript__1674__432325__319865__16.0.     1         3  15761 Ki    0.48%  15726 Ki 15847 Ki      63 ns    20%

Link to this section Summary


Defines text report format.

Link to this section Types

-type format_options() :: #{viewport_width => pos_integer(), format => basic | extended | full}.

Defines text report format.

  • format:
    • basic: default format containing only average throughput per sample_duration and average runner runtime
    • extended: includes median, p99 and other metrics(default for 10 and more samples)
    • full: includes system information in addition to extended output
    See overview for the detailed description
  • viewport_width: how wide the report can be, defaults to io:columns/0. Falls back to 80 when terminal does not report width.

Link to this section Functions

Link to this function

format(Reports, Options)

View Source
-spec format(Reports, Options) -> iolist()
          when Reports :: [erlperf:report()], Options :: format_options().

Formats result produced by erlperf:benchmark/3.

Requires full report. Does not accept basic or extended variants.