Module Bench

module Bench: sig .. end

Simple example:

open Core.Std module Bench = Core_extended.Bench

let main () = Bench.bench Bench.Test.create ~name:"test" (fun () -> ignore (Time.now ())) ;;

let () = main ()

Simple example:

open Core.Std module Bench = Core_extended.Bench

let main () = Bench.bench Bench.Test.create ~name:"test" (fun () -> ignore (Time.now ())) ;;

let () = main ()

module Test: sig .. end

module Result: sig .. end

verbosity (default low): If low, only prints results. If mid, additionally prints time estimates and a status line. If high, additionally prints information at each step of benchmarking.

gc_prefs: can be used to set custom GC settings during benchmarking (they will be reverted when the function returns)

no_compactions (default false): disables compactions during benchmarking, reverted when the function returns. Takes precedence over gc_prefs.

fast (default false): run fewer tests and thus get less accurate results in less time.

clock (default wall): controls time measurement method. Wall includes waiting on I/O and when the process is suspended/descheduled; cpu only counts time spent on computations.

type 'a with_benchmark_flags = ?verbosity:[ `High | `Low | `Mid ] ->
       ?gc_prefs:Core.Std.Gc.Control.t ->
       ?no_compactions:bool -> ?fast:bool -> ?clock:[ `Cpu | `Wall ] -> 'a

type 'a with_print_flags = ?time_format:[ `Auto | `Ms | `Ns | `S | `Us ] -> ?limit_width_to:int -> 'a

time_format (default auto): select the units to report run times in. If auto, the units are chosen based on the times.

The "Name" and "Input size" columns of the printed table reflect the values passed to Test.create. The "Normalized" column is run_time / input_size. "Stdev" reports the standard deviation for the "Run time" column. "Allocated" reports the average number of allocated words through the benchmarks.

"Warnings" may contain single characters indicating various things: 'm' indicates the minimum run time was less than 80% of the mean 'M' indicates the maximum run time was more than 120% of the mean 'c' indicates GC compactions occurred during testing 'a' indicates the number of words allocated was not the same in all tests

val bench : (Test.t list -> unit) with_benchmark_flags with_print_flags

val bench_raw : (Test.t list -> Result.t list) with_benchmark_flags

bench_raw returns a list documenting the runtimes rather than printing to stdout. These can be fed to print for results identical to calling bench.

val print : (Result.t list -> unit) with_print_flags