Richard Lynch wrote: > On Wed, April 25, 2007 5:48 am, Henning Eiben wrote: >> But as far as I understand, java kinda does the same thing, doesn't >> it? > > Sort of, but not really... > > Java has a single central one giant process architecture. > > PHP has a shared-nothing architecture. > > They're at polar opposites on this bit. > > A PHP accelerator/cache wedges the Java-ness of caching the oft-used > code in RAM, without the baggage of having everything in one giant > process. OK, but when I run PHP as an ISAPI filter, I see in my process-explorer (systinternals) only one process (w3wp.exe), that has a lot of threads. This must be the concurrent request being served. Comparing this to Java, I also have only one process (java.exe) and a huge amount of threads. I would assume, that creating java-threads within a JVM is more lightweight than creating new threads at the OS-level like PHP needs to do. >> Do you know any resource, describing how php works (like the thing >> about >> compiling opcode, and stuff)? How does php handle simultaneous >> requests? > > Woof. > > There are TONS of resources out there, most of which are probably > irrelevant to your actual problem, if you'd just profile the code and > find out what's actually going on. Well, I want to describe my observations I made with PHP and Java, and I want to explain why PHP is so much slower, compared to Java. So I would like to add some references to my paper ... and to be honest, I'm afraid I would run out of time if I would profile PHP to figure exactly out what's going on. So I don't need the source-code of PHP to show what's *exactly* going on, but roughly describing how PHP works. >> Are there multiple threads spawned? > > And in Windows, maybe, where the whole concept of process/thread is of > a different nature than Linux threads/processes. hmm, too bad I'm stuck with windows :) > PHP run under Apache lets Apache handle the process/thread splitting. Well, same with windows ... I would assume that the IIS is handling the whole process/thread stuff > Again, though, profile your code, please. hmm ... as mentioned above - but I might give it a try -- ... The UART's won't take this speed, Captain -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
