This is a recreation of the challenge 1brc proposed by Gunnar in Common Lisp. I am no expert and so this is probably not the best code. I have also ignored the "no external dependencies" rule because I needed to call mmap and babel (for utf8 encoding).

This runs in ~20 seconds in a Intel i7-9750H with 16GB of RAM. It has 12 cores so if you want to re-create it, you should change it according to your CPU.

• The original repo: https://github.com/gunnarmorling/1brc
• Python implementation: https://github.com/ifnesi/1brc
• A blog with cool notes on speed: https://blog.kingcons.io/posts/Going-Faster-with-Lisp.html

You can use make build or make run to do what it says. You can also load the project with quicklisp in a repl and hack away at it. To generate the data/measurements.txt you need to run createMeasurements.py 100_000_000. This will create a file with a billion (100.000.000) entries.