The benchmarking CLI has four main modes of operation:

1. Run all: yarn bench or yarn bench all
2. Run a specific suite: yarn bench suite Arr
3. Run a specific benchmark: yarn bench single Arr.foldr
4. List available suites and their benchmarks (JSON format): yarn bench list

The suite and single modes support a --sizes option, which accepts a comma-separated list of values (e.g. 10,20,30) as an arugment. Sized benchmarks will run once with each of the provided sizes, overriding the default sizes specified in the suite:

$ yarn bench single Arr.foldr --sizes 10,100,1000
Running benchmark Arr.foldr
Arr: foldr - 10 elements [...]
Arr: foldr - 100 elements [...]
Arr: foldr - 1000 elements [...]

The benchmark.js results for any of the modes can be saved in JSON format by using the --save-to option followed by a file path. The output file will also contain the UTC time, current commit SHA, current branch, and package.json version.

A benchmark is either sized or unsized. See ./util.js for more information.

A suite is an object:

type Suite a = 
  { name         :: String,
  , benchmarks   :: [Bench a],
  , defaultSizes :: [Int]
  }

See the exports of ./instances/arr.js and ./instances/obj.js for examples.

Any suite added to the list exported from ./instances/index.js will be available on the CLI.