On Tue, 1 Jun 2004 02:21 pm, Neal Richter wrote: > > The term "profiling" sounds like seeing where the CPU time is > > going. Now that Gilles has fixed the regex problem, CPU time is a > > factor of two or three less than elapsed time, so reducing CPU > > time won't really help. We need a way to "profile" the I/O. > > (Since the web I/O > > Profiling can do that.. it shows where time is being spend in > functions... and waiting on I/0 involves a system call.. which is > profiled.
Fair enough. I'm just thinking of what gprof gives, and that doesn't record the time that the process is waiting on I/O. It could count the *number* of system calls, but I don't know any way to find out how long each one takes. > Note that logic would indicate that 3.2.0 is likely NOT I/O > bound, since we aren't fundamentally doing much different as far as > I/O than in 3.1.x.... not enough to cause a 500% slowdown anyway. My only evidence is that the time reported by gprof is very much less than the elapsed time, and (from memory) is comparable with the time taken by 3.1.6. It is possible that it is only some combinations of attributes that cause the problem -- since Ted says he finds 3.2.0b5 faster -- or perhaps that I'm just not getting reliable results from gprof. (I tried looking for Joe's profiles, but the 3.2.0b5 directory in his patch site seems not to be working at the moment.) Another possibility is a bug in the code which pauses so as not to overload the web server -- that would also cause CPU usage to be much below elapsed time. Anyway, I agree with the general sentiment that we should release 3.2.0 before getting to the bottom of this, but with a suitable caveat. -- [EMAIL PROTECTED] ht://Dig developer DownUnder (http://www.htdig.org) ------------------------------------------------------- This SF.Net email is sponsored by: Oracle 10g Get certified on the hottest thing ever to hit the market... Oracle 10g. Take an Oracle 10g class now, and we'll give you the exam FREE. http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click _______________________________________________ ht://Dig Developer mailing list: [EMAIL PROTECTED] List information (subscribe/unsubscribe, etc.) https://lists.sourceforge.net/lists/listinfo/htdig-dev
