1BRC in Haskell: it's easy to get performance similar to 'wc'.

[hugodro]

Using the 1BRC to experiment with getting good performance out of Haskell normal code.

Conclusion: yes, Haskell code is performant almost out of the box. The trick is to take care of removing lazy evaluation where it's not useful.

I use the word-count utility wc as the benchmark for what is quick processing. In this case a 30 minute coding session resulted in a parser that runs 4-5x slower than wc, depending on the CPU/OS where the test is done.

The code is in: https://github.com/hugodro/fudd-1brc.

[jmatsushita]

For those interested in a fairly thorough treatment of the challenge, there's a discussion on the Haskell discourse with more implementations