Support the ongoing development of Laravel.io →
Octane Artisan

Specs:

  • Octane Version: 1.0
  • Laravel Version: 8.47.0
  • PHP Version: 8.0.7
  • Server & Version: Swoole 4.6.7
  • Specification Server: cpu 16 core, ram 32 GB, hdd 100 GB

I'm trying to benchmark when using php artisan and not, and when not using php artisan, i can get 3000+req/second using wrk.

  1. Benchmark using php artisan octane:start --server=swoole

result in another server to test using wrk wrk -t4 -c100 -d10s http://10.200.210.64/ image cpu tree in server octane using htop: The root process of swoole_http_server is below php artisan octane:start --server=swoole image CPU usage in server octane and all CPU not maximize, each cpu approximately 30%: image

  1. Benchmark without php artisan:

result in another server to test using wrk wrk -t4 -c100 -d10s http://10.200.210.64/ image the CPU tree in server octane using htop: This root process swoole_http_server not using php artisan image CPU usage in server octane and all CPU is push to max usage, each cpu approximately 66%+: image

  1. Conclussion: When not using php artisan i get 3 times faster than php artisan

Steps To Reproduce:

How to reproduce running without php artisan, actually im using php artisan to run this, but this method is tricky first i run php artisan in background using php artisan octane:start --server=swoole & then i close the ssh terminal and finally i can running octane without using php artisan and maximize cpu usage

is this bug or some feature? i hope i can maximize the cpu usage when using php artisan.. nb: i try adding some flag like --workers=16 and --task-workers=16 the result is same not improve cpu usage.

Last updated 3 years ago.

mysic liked this thread

1

Sign in to participate in this thread!

PHPverse

Your banner here too?

Moderators

We'd like to thank these amazing companies for supporting us

Your logo here?

Laravel.io

The Laravel portal for problem solving, knowledge sharing and community building.

© 2025 Laravel.io - All rights reserved.