Guy Harris wrote:
>
>
> The two changes in question cut the CPU time spent reading in the
> capture file in question from 36.12 seconds to 3.25 seconds - and the
> loop that reads in that list appears to be CPU-bound (my disk makes
> little noise, probably because the entire file fits in memory, and
> remained cached); the wall-clock time in a test (somewhat crude,
> went from 35 seconds to about 4 seconds.
>
My own stress-test shows that (Linux PII 350) :
From: 208.98user 0.81system 3:38.89elapsed 95%CPU
To: 3.88user 0.42system 0:14.49elapsed 29%CPU
That sounds like a really great performance improvement.
--
Laurent DENIEL | E-mail: deniel@xxxxxxxxxxx
Paris, FRANCE | laurent.deniel@xxxxxxxxxxxxxxxxxxxxxxxxxxxx
| WWW : http://www.worldnet.fr/~deniel
All above opinions are personal, unless stated otherwise.
Powered by MHonArc 2.6.10