1 Billion Row Challenge in PHP

This is my solution for the 1 billion rows challenge (https://1brc.dev), in PHP.

Create measurements.txt file :

php createMeasurements.php {number of reccords}

If you do have php installed locally but have docker, use ./phpd instead of php to run the command dockerized.

Run with Xdebug or the JIT

Build the docker image that contains the two extensions: docker build --tag "1brc" .. This builds the 1brc:latest image locally.

Use the phpd script by passing it PHP configuration directives with the -d option.

Ie:

Profile with Xdebug: ./phpd -d="xdebug.mode=profile" calculateAverage_v4.php 1M 5
Enable the JIT: ./phpd -d="xdebug.mode=off" -d="opcache.enable_cli=1" -d="opcache.jit_buffer_size=100M" calculateAverage_v4.php 10M 5

The script run via the php:8.3-cli docker image on Ubuntu 24.04 that itself run in a VM on Windows 10.
CPU speed is 3.5Ghz per real core.

version	Description	Total time	lines / ms	Peak memory usage
1	The first, naive version	112s for 100M rows	893	792 Kb
2	Replace usage of `fgetcsv` by `fgets`	37s for 100M rows	2660	792 Kb
3	Optimize the body of the loop that reads the data to have less array access	25s for 100M rows	4000	804 Kb
4	Use a object instead of array to hold the data	23s for 100M rows	4250	684 Kb
4.1	Same script as above but with the JIT enabled	19s for 100M rows	5250	621 Kb