README.md

In bjmt/autobenchR: Automated code benchmarking

autobenchR

Easily write and run large sets of benchmarks all at once, and collect results in one output file. The autobenchR package does not provide any benchmarking functionality by itself, instead allowing for use of one of the following packages: bench, microbenchmark, and rbenchmark.

Installation

if (!requireNamespace("remotes", quietly = TRUE)) install.packages("remotes")
remotes::install_github("bjmt/autobenchR")

Usage

Code

autobenchR::begin("results.txt", tool = "bench")

autobenchR::run("Tests",

  e1 = runif(100000, 0, 10),
  e2 = rnorm(100000, 5, 2.5)

)

autobenchR::update(tool = "microbenchmark")
autobenchR::run("Different tool",

  e1 = runif(100000, 0, 10),
  e2 = rnorm(100000, 5, 2.5)

)

autobenchR::update(tool = "rbenchmark", unit = "ms")
autobenchR::run("Different tool 2",

  e1 = runif(100000, 0, 10),
  e2 = rnorm(100000, 5, 2.5)

)

autobenchR::skip()
autobenchR::run("Skipped tests",

  e1 = sample(1:1000, 1000)

)

autobenchR::note(

  "* Bullet point 1",
  "* Bullet point 2",

  name = "Some notes"

)

autobenchR::run("Failing test",

  e1 = 1 + "a"

)

autobenchR::end()

Terminal output

Starting benchmarks
* Running benchmark 1: Tests [0.9 s]
* Running benchmark 2: Different tool [1.5 s]
* Running benchmark 3: Different tool 2 [1.7 s]
* Running benchmark 4: Skipped tests [SKIPPED]
* Running benchmark 6: Failing test [ERROR]
All benchmarks completed.
Total runtime: 4.1 seconds

Text results

See example-text.txt for the output file, and example-markdown.md for the markdown version.

bjmt/autobenchR documentation built on May 7, 2019, 12:51 a.m.