Skip to content

hamadmarri/elixir-benchmark-vs-other-languages

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Elixir Benchmark vs other languages

The test is a Depth First Search DFS on order of integer numbers from 1 to 10. Simply push to the stack the next generation of numbers and pull. This method tests the performance of a language on the following: -

  • Loop speed
  • Pushing and pulling from a list / stack
  • Check equality of two lists / arrays

The entropy is 10 numbers, and the maximum digits is 10. The current code is after some optimization. First Elixir was faster than Go about 3x, until I changed the default slice capacity in Go. After that Go become faster by 2x. So, it seems that the Go's slice initial capacity has much effects in performance.

Anyways, with the final optimizations on Elixir, Go, and Java, Elixir in this test is faster than Java by about 2x, and slower than Go by 1.2x

It is quite surprizing how Elixir is comparable with a complied language such as Go. Next I am planning to test C++ or C to see how far we are from almost optimum speed.

Results

  • Go average: 17.28828s ~17.3s
  • Elixir average: 21049749us ~21s / 3.75s / 1.22x slower than Go
  • Java average: 41.5353454s ~41.5s - 2.4x slower than Go and 1.97x slower than Elixir

Summary

Even thought some benchmarks on the Internet shows that Elixir might be not fast on computation or calculation tasks, Elixir in the DFS test is almost as a compiled language. Notice that the test DFS in this example is simple. I like Elixir as a beautiful Functional Programming language and the fact that it is based on Erlang technology. I was wishing before implementing GO that Elixir at least be slower no more than 4x or at least a linear (not exponential) differences. I am surprized that Elixir / Erlang at least in this test shows that it is not a slow languages as usually shown in different benchmarks in the Internet.

You may clone this repo and test. Please see the full results, build / run commands in results.txt files. If you have any optimized modification on either language please do not hesitate to Pull Request. Moreover, if you like to share your benchmarks on your machine I will be happy to share it as well as a separate branch such as bench-yourname.

Thank you

Hamad