tweag/hyperion

Name: hyperion

Owner: Tweag I/O

Description: A lab for future Criterion features.

Created: 2016-12-27 14:16:44.0

Updated: 2018-03-16 15:05:52.0

Pushed: 2018-01-08 08:09:14.0

Homepage:

Size: 104

Language: Haskell

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

Hyperion: Haskell-based systems benchmarking

Hyperion is a DSL for writing benchmarks to measure and analyze software performance. It is a lab for future Criterion features.

Getting started

Build

You can build the micro benchmark example using stack:

ack build
ack exec hyperion-micro-benchmark-example

Example usage

The Hyperion DSL is a backwards compatible extension to Criterion's DSL (except for the rarely used env combinator, which has a safer type). Here is an example:

hmarks :: [Benchmark]
hmarks =
[ bench "id" (nf id ())
, series [0,5..20] $ \n ->
    bgroup "pure-functions"
      [ bench "fact" (nf fact n)
      , bench "fib" (nf fib n)
      ]
, series [1..4] $ \n ->
    series [1..n] $ \k ->
      bench "n choose k" $ nf (uncurry choose) (n, k)
]

 :: IO ()
 = defaultMain "hyperion-example-micro-benchmarks" benchmarks

By default Hyperion runs your benchmarks and pretty prints the results. There are several command-line options that you can pass to the executable, like printing the results to a JSON file or including individual raw measurements. To see the full set of options run the executable with --help:

ack exec hyperion-micro-benchmark-example -- --help
e: hyperion-micro-benchmark-example ([--pretty] | [-j|--json PATH] |
                                    [-f|--flat PATH]) ([-l|--list] | [--run]
                                    | [--no-analyze]) [--raw]
                                    [--arg KEY:VAL] [NAME...]

lable options:
,--help                Show this help text
pretty                 Pretty prints the measurements on stdout.
,--json PATH           Where to write the json benchmarks output. Can be a
                       file name, a directory name or '-' for stdout.
,--flat PATH           Where to write the json benchmarks output. Can be a
                       file name, a directory name or '-' for stdout.
version                Display version information
,--list                List benchmark names
run                    Run benchmarks and analyze them (default)
no-analyze             Only run the benchmarks
raw                    Include raw measurement data in report.
arg KEY:VAL            Extra metadata to include in the report, in the
                       format key:value.