Which programs are fastest? Le mieux est l'ennemi du bien.

Do the Time-used boxes overlap or is there clear separation between them? Which programming language implementations have the fastest programs?

  language implementation  |-  |---  25% median  75%  ---|  -|

GNU gcc 1.001.001.001.001.301.749.82

C++ GNU g++ 1.001.001.001.061.141.364.83

ATS 1.001.001.061.293.305.855.85

Ada 2005 GNAT 1.091.091.461.683.045.428.64

Java 6 steady state 1.051.051.381.682.544.2811.33

Java 6 -server 1.061.061.441.983.005.3511.45

Scala 1.031.032.052.805.3510.3221.33

Haskell GHC 1.341.342.263.005.369.479.47

C# Mono 1.331.332.123.3311.5525.7051.01

OCaml 1.931.932.563.637.7515.54322.91

Lisp SBCL 1.511.512.414.175.5710.3127.45

Pascal Free Pascal 1.491.492.434.566.2612.0114.14

Lua LuaJIT 1.421.423.125.2312.0425.4329.87

Clean 1.221.223.875.698.2914.9116.90

Fortran Intel 1.021.021.636.7411.2625.0525.05

F# Mono 2.122.123.928.1831.3249.3449.34

Go 6g 8g 3.643.648.0713.5728.6448.6548.65

Erlang HiPE 3.783.787.8616.6425.4351.78154.62

Scheme PLT 2.072.0711.9122.2772.45163.27310.68

Java 6 -Xint 2.032.0312.5623.4227.3949.6454.09

JavaScript V8 1.001.008.4927.0756.4186.5786.57

Smalltalk VisualWorks 9.089.0820.9733.1756.09108.77196.14

Lua 1.431.4322.9234.8487.41156.09156.09

PHP 2.812.818.3851.7597.77129.17129.17

Python 3 5.735.7335.9156.79182.07379.48379.48

Python CPython 2.732.7318.4558.18119.88272.02338.19

Mozart/Oz 10.3910.3934.9760.10172.13377.89397.79

JavaScript TraceMonkey 1.511.5128.8166.84114.92244.081591.64

Perl 2.522.526.2973.43101.54244.41259.61

Ruby JRuby 14.9714.9724.8476.27146.08298.98298.98

Ruby 1.9 10.0410.0436.6288.06220.49496.311156.05

Ruby MRI 10.3510.3570.65169.30368.12814.332113.29

 Which programs are fastest?


Which programming language implementations have the fastest programs?


Read "Visual Presentation of Data by Means of Box Plots" (pdf)

For each benchmark, F is the fastest time measurement; for each language implementation, each time measurement T is then normalized to T/F


There are other ways to analyse and present this data: look for patterns in Code-used Time-used Shapes or ask Which is best for you? or take the Summary Data and do your own analysis!

Revised BSD license