Skip to content

a benchmark for compile-time and/or runtime Nim πŸ†

License

Notifications You must be signed in to change notification settings

disruptek/golden

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

golden

A benchmarking tool that measures and records runtime of any executable and also happens to know how to compile Nim.

  • cpp +/ nim-1.0 Build Status
  • arc +/ cpp +/ nim-1.3 Build Status

The idea here is that we're gonna make a record of everything we run, everything we build, and be able to discover and pinpoint regressions automatically using native git-fu performed by the tool. Fire and forget!

Installation

Nimph

$ nimph clone golden

Nimble

$ nimble install golden

Usage

If you pass it a binary, it'll run it a bunch of times and report some runtime statistics periodically.

If you pass it some Nim source, it will compile it for you and report some compilation and runtime statistics periodically.

By default, it will run until you interrupt it.

$ golden --truth=0.002 bench.nim
compilations after 0s
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Builds β”‚ Min      β”‚ Max      β”‚ Mean     β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      1 β”‚ 0.396129 β”‚ 0.396129 β”‚ 0.396129 β”‚ 0.000000 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
benchmark after 1s
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Runs   β”‚ Min      β”‚ Max      β”‚ Mean     β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      1 β”‚ 1.959187 β”‚ 1.959187 β”‚ 1.959187 β”‚ 0.000000 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
benchmark after 3s
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Runs   β”‚ Min      β”‚ Max      β”‚ Mean     β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      2 β”‚ 1.958892 β”‚ 1.959187 β”‚ 1.959039 β”‚ 0.000147 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
completed benchmark after 5s
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Runs   β”‚ Min      β”‚ Max      β”‚ Mean     β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      3 β”‚ 1.958892 β”‚ 1.961293 β”‚ 1.959791 β”‚ 0.001069 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Benchmarking the compilation of Nim itself:

$ cd ~/git/Nim
$ golden koch -- boot -d:danger
# ...
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ #      β”‚ Min      β”‚ Max      β”‚ Mean     β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚     12 β”‚ 8.846606 β”‚ 9.485832 β”‚ 8.945023 β”‚ 0.165638 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Benchmarking compilation of slow-to-compile Nim:

$ golden --compilation openapi.nim
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚      # β”‚ Min       β”‚ Max       β”‚ Mean      β”‚ StdDev   β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      1 β”‚ 91.946370 β”‚ 91.946370 β”‚ 91.946370 β”‚ 0.000000 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”¬β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚      # β”‚ Min       β”‚ Max       β”‚ Mean      β”‚ StdDev    β”‚
β”œβ”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€
β”‚      2 β”‚ 29.271556 β”‚ 91.946370 β”‚ 60.608963 β”‚ 31.337407 β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Command Line Options

  • truth a float percentage indicating how much jitter you'll accept
  • runtime a float of seconds after which we should kill each invocation
  • iterations a number of invocations after which we should stop the benchmark
  • storage the path to a database file you wish to use; must end in .golden-lmdb
  • interactive-forced assume output friendly to humans
  • json-output assume output friendly to machines (work in progress)
  • color-forced enable color output when not in interactive mode
  • prune-outliers throw out this percentage of aberrant invocations with long runtime in order to clean up the histogram
  • dry-run don't write any results to the database
  • histogram-classes the number of points in the histogram
  • compilation-only benchmark the Nim compiler on the given source(s)
  • brief only output the statistics at the completion of the benchmark
  • never-output never emit anything via stdout/stderr
  • dump-output always print the stdout/stderr of the benchmarked program
  • -- the following arguments are passed to the compiler and runtime. Note that if you supply -- cpp for compilation via C++, you will need to supply your own defines such as -d:danger.

License

MIT