Re: CVS update: php3/functions

At 09:03 27/12/98 +0100, Sascha Schumann wrote: >On Sun, 27 Dec 1998, Rasmus Lerdorf wrote: > >> > > How much, can you provide some figures or benchmarks? >> > >> > I just ran a short benchmark (regal.c) here which showed that at 200000 >> > iterations we save about one third of user time. Attached is a short graph >> > which shows the results of two tests with and without the cache. >> >> It should be noted that this improvement is only seen when the same >> regular expression is applied repeatedly. An application with a bunch of >> different regex calls is not going to speed up. > >>From the log: > >revision 1.89 >date: 1998/12/25 20:09:47; author: sas; state: Exp; lines: +46 -17 >speed up regexes by introducing a simple cache which is >effective in the simplest case - using always the same regex How costly are regex's? If they're extremely costly, we can have a bigger cache, implemented as a limited-size hash table implemented in FIFO. A lookup would be fairly quick (basically, a hash lookup with the pattern being the key) so if regex's are relatively slow (and aren't O(1), which they aren't) it would give a noticable performance boost. That table can also be persistent (i.e., saved across hits) which would increase performance even more. Zeev -- Zeev Suraski <zeev@zend.com> For a PGP public key, finger bourbon@netvision.net.il -- PHP Development Mailing List http://www.php.net/ To unsubscribe send an empty message to php-dev-unsubscribe@lists.php.net For help: php-dev-help@lists.php.net