https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6060





--- Comment #9 from Justin Mason <[email protected]>  2009-04-01 03:01:56 PST ---
> Whatever speedup there was must have been due to different
> memory organization of compiled code, I can't see other reason.

yeah.  I have a feeling that as CPUs have gotten so fast compared to
the memory pipeline, the effects of memory usage reduction become 
more important for execution speed.

> > can you attach the other changes as patches and I'll test them too?

they don't look so great:

BEFORE
time ./mass-check -n -o ham:mbox:/home/jmason/ftp/tstcor/ham.mbox 
spam:mbox:/home/jmason/ftp/tstcor/spam.mbox > /dev/null
real    0m32.587s user    0m32.166s sys     0m0.284s
real    0m32.711s user    0m32.266s sys     0m0.264s
real    0m32.416s user    0m32.002s sys     0m0.340s
real    0m32.450s user    0m32.094s sys     0m0.276s

real   avg=33.956s min=33.537s max=34.401s count=5
user   avg=33.304s min=32.986s max=33.834s count=5
sys    avg=0.282s min=0.236s max=0.312s count=5


AFTER
time ./mass-check -n -o ham:mbox:/home/jmason/ftp/tstcor/ham.mbox 
spam:mbox:/home/jmason/ftp/tstcor/spam.mbox > /dev/null
real    0m33.833s user    0m33.078s sys     0m0.308s
real    0m34.401s user    0m33.834s sys     0m0.296s
real    0m33.913s user    0m33.266s sys     0m0.312s
real    0m34.098s user    0m33.358s sys     0m0.236s

real   avg=32.501s min=32.269s max=32.711s count=6
user   avg=32.111s min=31.902s max=32.266s count=6
sys    avg=0.282s min=0.224s max=0.340s count=6

So that's 4.4% slower, it seems.  (now this wasn't on an unloaded machine, but
I think those numbers are mostly accurate.)

I'm afraid I'm -1 on that with that performance reduction; sorry!


-- 
Configure bugmail: 
https://issues.apache.org/SpamAssassin/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

Reply via email to