Here are preliminary results comparing php56 vs php70: https://github.com/motin/php-framework-benchmark/blob/multiple-stacks/OTHER_RESULTS.md#motin

Interesting!

Do the zeros represent failed tests?

The No-framework number changed little between 5.6 and 7.0. Is that because that case is bound by something other than PHP script execution, e.g. fpm?

@motin How many times do you run the benchmarks?
Did you confirm the results are roughly the same all the time?

I don't understand why tipsy is faster than phalcon, codeigniter is so slow, on PHP-FPM 5.6.4 with opcode cache.

Do the zeros represent failed tests?

Yes indeed

The No-framework number changed little between 5.6 and 7.0. Is that because that case is bound by something other than PHP script execution, e.g. fpm?

The environments are identical in set-up, hardware, OS and configuration except for the fact that php7.0-fpm is used instead of php5-fpm. I'd say it is pretty fair to assume that PHP 7 is simply a bit faster even in the hello world case.

How many times do you run the benchmarks?
Did you confirm the results are roughly the same all the time?

I ran them a couple more times, and the results vary quite a bit. Sometimes tipsy is twice as fast as phalcon. Will run this in production environment and report back the results.

I don't understand why tipsy is faster than phalcon, codeigniter is so slow, on PHP-FPM 5.6.4 with opcode cache.

Haven't checked, but maybe because it is extremely light-weight and more similar to a the no-framework set-up than a full-fledged framework.

It's good idea to add column for php5.6 and another column for php7 in final results picture.

Also please add ZF3