The time given by the browser is likely more accurate.

I'm not sure how the debug bar measures the request time but if it is in the php process there will be at least some time that passes before the timer is started and some time after it is stopped.

As for the opcache being slower if the files used in the request are not yet cached it will be slightly slower as it has to save the compiled opcodes in the cache as well as the normal stuff it does. Once they are cached it should be faster though.

spekkionu: Most of the time you use a library like ChromePHP and creates some big headers that go out and the toolbar reads it.

Well I think it is related as you said spekkionu with discrepancy of script start and end timestoo, as well as apache maybe? Will somebody share their apps performance? :)

I wanted to dev using laravel and I was amazed that it took 80-100ms waiting time on a localhost using a simple hello world, I decided not to use it in production since a normal php takes 3-5 ms

Ok, old thread, but I'm working through some of this myself right now. Keep in mind, the total wait time from the browser is including all of the tcpip overhead and latency between you and your server. That alone could represent the discrepancies you are seeing between debugbar and browser tools.

Klaniak, while you make a reasonable point, keep in mind, a hellow world test and a real application are two very different beasts. Raw code will always win out. That said, there are certain production ready optimizations that can be performed that significantly speed up response times of even hello world.