time test

[David Relson]

Matthias,

My curiosity was aroused by your observation that bogofilter performance 
has dropped recently, as measured by processing the 900k file generated by 
your perl script.

I ran 0.9.1.2, 0.10.2, 0.10.3.1, current cvs, and my current tst version on 
your 900k file and Greg's 600k file, i.e. 4.txt.  As you can see, 
bogofilter's performance on the 900k file is much better than on the 600k file:

         300k*"ABC"	   4.txt
0912	0.01user	  0.13user
01021	4.48user	150.05user
01031	4.54user	149.74user
cvs	4.55user	148.85user
tst	4.66user	  0.23user

FWIW, I've also done some tests with compiler optimization.  Taking out the 
"-O2" to build a debug version creates a bogofilter that takes about 3 
times as long to perform any of the tests.  For 0.10.2.1 times are 12.15s 
instead of 4.48s and 425.08s instead of 150.50s.

David