Re: Benchmarking PHPA and Cache_Lite

On Wed, 12 Feb 2003, Fabien MARTY wrote: > I think that Cache_Lite can be really faster. If you have a look at the > Cache_Lite code (or technicals docs), you will note that I don't include > PEAR.php for speed reason ! Indeed, the inclusion is dynamic (with > include_once and not require_once) when an error occured but PEAR.php > isn't included in most cases. With this system, the speed jump is very > important. You're quite right. Moving the requires as an include inside a conditional block and icluding extra stuff only when necessary doubled performance even more. ab -n 1000 -v 3 -c 10 http://www.example.com/test/select_1000_cached2.php new results (original numbers in braces) Time taken for tests: 15.295 seconds (30.357 seconds) Requests per second: 65.38 (32.94) Transfer rate: 4351.54 kb/s received (2195.52 kb/s) -- Mika Tuupola http://www.appelsiini.net/~tuupola/