> For example your quote: NekoVM is around 30 times faster than PHP.. In > all honestly I kind of doubt it, since this is coming from you.. The > first thing that pops is if you installed an accelerator or not.. This > is something every PHP developer should do in a production enviroment.. > For all I know you installed PHP as a CGI.. this is something some of > the *AMP installers for windows do by default.. > > If you have the numbers I will take everything back =)
This is some numbers when running the binary-trees benchmark (from the great computer language shootout), with N=14 . I'm using PHP 5.1.4 Win32 binaries and Neko CVS Win32. - neko : 3.5 seconds - php : 1 minute 40 seconds That's an 28 times faster, so not exactly 30, but Neko has additional optimizations with GCC so perform on an average 20% faster on Linux/OSX, which is not taken into account in this comparison. You might want to run your own tests to check about it. For more basic benchmarks such as fibonnacci numbers, then you get "only" a 6 times faster than PHP. But the binary-tree covers more VM features than fib does. I'm talking here about raw VM computation power. PHP accelerators are only doing bytecode caching which does not change anything for these cases since the code is really small and compiling it does not account for much of the time spent computing. Neko has bytecode caching as well but a bit more powerful since your static variables are kept initializated between two runs, enabling you for example to keep an already parsed XML document into Apache memory. Nicolas _______________________________________________ osflash mailing list [email protected] http://osflash.org/mailman/listinfo/osflash_osflash.org
