On Tue, Jan 29, 2002 at 09:23:56PM -0500, dman wrote: > On Tue, Jan 29, 2002 at 08:39:16PM -0500, Duncan Findlay wrote: > | On Tue, Jan 29, 2002 at 07:13:09PM -0500, dman wrote: > | > Sure. And a default to off of any superfluous tests achieves this as > | > well. BTW, Duncan, are you using spamd? On my (relatively fast) > | > system I see a major difference bewteen using spamd and > | > 'spamassassin'. (most scans take 1 second or so with spamd on my box) > | > > | > | Yes. I am. I think it would take an hour after I start my computer is I > | used spamassassin -p, rather than 20 minutes :-) > > Wow. I wonder what the cause is ... probably CPU due to heavy regex > usage. I wonder if it would be more efficient to iterate over > the message for each test or to run each test in parallel with just > one iteration over the message. The second would surely require more > memory to keep track of each test. Profiling is the only way to be > sure. In the current implementation, does SA stop checking a test > when it reaches the first match? That may improve the processing time > on spam messages. > > -D >
I'm guessing RAM is the limiting factor. It's a PI/100 w/ 40 MB RAM. -- Duncan Findlay _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
