Web Framework Benchmarks at TechEmpower

Does anyone want to submit Phalcon for these:

http://www.techempower.com/blog/2013/03/28/framework-benchmarks/

https://github.com/TechEmpower/FrameworkBenchmarks

It seems CakePHP didn't fare too well, and even plain old PHP had some performance issues compared to NodeJS and JVM based competitors. Would be interesting to see how it fares in comparison.

That test looks weird. Nothing good comes out when comparing square to a salad.



5.3k

There is nothing weird in it - you should know for what to use the square and for what the salad. Isn't the performance one of the main reasons why we are not using Zend or Symfony framework - because they are performing "as good" as CakePHP? And PHP (without ORMs like ActiveRecord or Doctrine) is performing exceptionally well for database operations!

About the PHP performance issues - this is the tradeof for using high level language and not the low level one like C or Java. We have made the choice to use PHP so we shouldn't complain about it :). I think it is the right choice for web development and also PHP performance is getting better with each new version (the author of the test used PHP 5.3).



82.2k

I agree with @roman-kulish, an event-driven (asynchronous, non-blocking) framework cannot be compared with PHP (blocking, synchronous). The best benchmark the PHP world can offer is the raw PHP which is missing an ORM/Framework and multiple layers that try to make the software more structured and maintainable.

We're using PHP because it helps us to create robust web applications with rich frontends, api backends, etc. with almost the same syntax/framework without no-knowledge of network stuff.

Comparing Phalcon/Cake/Zend/Rails or any other synchronous framework with a network framework like Netty is useless.

A network oriented framework can be compared with Twisted, Tornado or Nodejs. There are no middleware intermediates like Apache/Nginx when using these frameworks so they can offer better performance but you need more effort to make them work like PHP.

Looks like someone added Phalcon to the latest round:

http://www.techempower.com/benchmarks/

edited Jul '16

Why is PHP5 ahead of PHP(7?) in this benchmark? That test looks weird indeed