SKIP timing
David Relson
relson at osagesoftware.com
Tue Nov 30 03:44:17 CET 2004
Matthias,
Using Evgeny's test message
-rw-r--r-- 1 relson relson 456505 Nov 26 18:32 48220.msg
I ran 3 timings:
1 - 0.93.1's standard parser
2 - using "if (dest!=source) memmove(dest,source,count)" instead of
"for (...) *dest++ = *source++"
3 - with SKIP code (as outlined earlier)
the times are:
1 - 26.94 user, 0.01 system, 31.69 elapsed, 85% CPU
2 - 23.16 user, 0.01 system, 28.73 elapsed, 80% CPU
3 - 12.54 user, 0.00 system, 12.84 elapsed, 97% CPU
Looks like further testing of the SKIP idea is worth doing.
By the way, "make check" isn't affected by my patch (which I've attached
for your experimentation).
Regards,
David
More information about the bogofilter-dev
mailing list