On Thu, April 26, 2007 5:53 am, Robin Vickery wrote: > On 25/04/07, Richard Lynch <[EMAIL PROTECTED]> wrote: >> On Wed, April 25, 2007 5:09 am, Zoltán Németh wrote: >> > 2007. 04. 25, szerda keltezÃ(c)ssel 11.53-kor Henning Eiben ezt >> Ãrta: >> >> Zoltán NÃ(c)meth schrieb: >> >> >> > not exactly. it pre-compiles them to opcodes and stores the opcode >> > blocks. the interpreter normally first pre-compiles the code to >> > opcodes >> > then runs the opcode. this pre-compilation can be cached with >> > accelerators, that's how they increase performance. >> >> Please, please, please stop misinformation. :-) >> >> The accelerators *ALL* make their dramatic performance boost by >> CACHING the hard drive into RAM. >> >> Caching PHP Source would do *almost* as well. > > This statement seems a bit suspicious to me - what OS are you using > that doesn't already cache the disk accesses into RAM? (for example > the Linux VFS buffer cache).
All the OSes have some kind of general-purpose cache for whatever files, all of them, that you've been using... That does not negate the statements I've made: The PHP specific accelerators cache the bytecode in RAM, which provides far more of a boost than bypassing that compile phase. Feel free to use callgrind on APC and see for yourself how it changes the behaviour of the system, and where the big "win" in performance comes from, if you think I'm wrong. I'm only telling you what I was told by Ze'ev, Andi, Rasmus etc, who actually wrote the dang things. :-) -- Some people have a "gift" link here. Know what I want? I want you to buy a CD from some indie artist. http://cdbaby.com/browse/from/lynch Yeah, I get a buck. So? -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php