Karol Grecki wrote: > > You have no way to tell how this choice affected the benchmarks, you > cannot assume that every framework was slowed down equally. For example if > I/O was particularly slow, that would penalize frameworks with larger > number of includes more than others, skewing the results. > We are talking about existing operating systems and enviroments, not about ideal, not existing (or experimental) enviroments with artificial ultra bad performing I/O management.
On existing enviroments, you'll find that the variations between Windows Vista + Apache performances and (GNU/)Linux + Apache performances won't compensate so much the "magnitude" of performance differences between slower and faster PHP scripts. We are talking about Linux + Apache and Windows + Apache, not Linux + Apache and an abacus. However, maybe you should develop that application avoiding over-engineering and using less PHP includes, because they are always not-so-fast (especially if you don't use bytecode caching systems) also on machines with the most performant I/O subsystems. > You think it's just about having the environment constant, would you be > able to tell which off-road car is better after testing both of the on the > same highway? No. > You should know these are computers and not cars (where you have aerodynamics issues etc.), your example is just plain wrong ;-) -- View this message in context: http://www.nabble.com/Framework-speed-shotout----question-tp19914787p20303803.html Sent from the Zend Framework mailing list archive at Nabble.com.
