Re: bug in svn version
"Jonathan Ellis" <[email protected]>
| Newsgroups | gmane.comp.python.spyce.general |
|---|---|
| Message-ID | <[email protected]> |
On Fri, 21 Apr 2006 09:55:45 +0200, "Michele Petrazzo" <[email protected]> said: > >> You can find them here (ok this are in Italian language, but there > >> are only numbers... If you want I can translate it ) > >> > >> http://tinyurl.com/fxrss http://tinyurl.com/e8ll2 > > > > Well, nobody *wants* to be slow, but here if I'm reading this right > > that just means you had ~100 req/s with Spyce vs ~1000/s with PHP, > > for instance. > > Yes and no, because on my Athlon(tm) XP 2600+, with "fork" I have: > > Requests per second: 6.83 [#/sec] (mean) > Time per request: 7319.638 [ms] (mean) > Time per request: 146.393 [ms] (mean, across all concurrent > requests) > Transfer rate: 55.95 [Kbytes/sec] received Ah, I see. That's the problem with not understanding Italian. :) I guess an easy way to test would be to write a script that just does 100 forks, 100 times in a row or so (but not 10000 forks at once, the scheduler won't like that. :) Then do something similar with threads and see if fork really is that much slower. In any case, I've removed forking as a concurrency option and switched threading to a pooled architecture. (Before it just used SocketServer.TCPServer which spawns a n ew thread for each request.) I'd be curious to see how your numbers compare with the new code. (Make sure you've switched to the sourceforge repository: you should be at revision 850 for this.) -J -- C++ is history repeated as tragedy. Java is history repeated as farce. --Scott McKay ------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642