0.18 performance results linux-headers-2.6.21-2 / 2752 files / PII
Jari Aalto
jari.aalto at cante.net
Thu Aug 2 13:34:51 BST 2007
* Thu 2007-08-02 Jari Aalto <jari.aalto AT cante.net>
> Here Are some performance results from PII 500 Mhz (Debian
> 2.6.21-2-686) / Ext3 => importing linux-headers-2.6.21-2 / 2752 files
Correction: These are 0.17 resuls
> $ time bzr init
> real 0m1.049s
> user 0m0.672s
> sys 0m0.108s
>
> $ time bzr add
> real 0m9.621s
> user 0m4.388s
> sys 0m0.316s
>
> $ bzr ci -m "- insert"
> real 2m4.805s
> user 0m42.919s
> sys 0m2.692s
>
> Total around 2min 15sec.
> I suspect that if bzr would reduce amount of information printed to
> the screen the times would be faster.
Indeed. Here are results from 0.18 (real this time):
$ bzr init
real 0m1.101s
user 0m0.620s
sys 0m0.128s
$ bzr add
real 0m6.219s
user 0m4.484s
sys 0m0.320s
$ bzr ci -m "- commit"
real 1m42.505s
user 0m43.171s
sys 0m3.596s
$ bzr ci -m "- commit" > /dev/null 2>&1
real 0m50.275s
user 0m41.455s
sys 0m2.804s
So, if there is no output the speed is under ONE minute! That'd be
over 200% improvement compared to 0.17
Should the messages be printed only if user requested -v or (multiples
of it like with ssh).
Jari
More information about the bazaar
mailing list