You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Mix <mi...@acm.org> on 2002/08/12 01:02:44 UTC
cvs/svn checkout/update timing info.
Here is a cvs/svn checkout timing comparison for you. Tests were
performed on a Sun Ultra 10 running Solaris 2.8. Both the cvs and svn
repositories were on a local hard drive. The working copies checked out
from each were to a NFS mount. Used ra_local. I ran the tests multiple
times and the results were within a few percent of each other so only
one result is posted for each. The source code in the repositories was
identical and was the only revision (import to fresh repos). Here are
the numbers:
CVS:
real 0m27.215s
user 0m0.240s
sys 0m0.760s
SVN (0.14.1):
real 5m8.038s
user 0m12.900s
sys 0m20.930s
SVN (2945):
real 4m55.424s
user 0m13.030s
sys 0m30.990s
Kind of sad huh :)
So then I went into each WC and performed an 'update' without having
changed a single thing. Here are the times:
CVS:
real 0m1.320s
user 0m0.030s
sys 0m0.200s
SVN (2945):
real 0m4.936s
user 0m0.210s
sys 0m0.400s
Here are the times for a 'status' (and yes I know they aren't identical):
CVS (output sent to /dev/null):
real 0m0.309s
user 0m0.100s
sys 0m0.110s
SVN (2945):
real 0m4.960s
user 0m2.810s
sys 0m0.570s
Just letting you know.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: cvs/svn checkout/update timing info.
Posted by Branko Čibej <br...@xbc.nu>.
Mix wrote:
> Here is a cvs/svn checkout timing comparison for you. Tests were
> performed on a Sun Ultra 10 running Solaris 2.8. Both the cvs and svn
> repositories were on a local hard drive. The working copies checked
> out from each were to a NFS mount. Used ra_local. I ran the tests
> multiple times and the results were within a few percent of each other
> so only one result is posted for each. The source code in the
> repositories was identical and was the only revision (import to fresh
> repos). Here are the numbers:
[snip]
Yup, all of those differences are probbaly related to our hopeless
management of the .svn/entries file.
> Just letting you know.
I have two horribly slow disks fighting for breath on the same IDE
channel; I'm very painfully aware of this already. :-) (Or should I say
:-( ?)
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: cvs/svn checkout/update timing info.
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Michael Price <mp...@atl.lmco.com> writes:
> I just did a 'echo "new file contents" > main.cc' in a cvs and svn WC and
> then timed the commits. Same test setup as previous Mix email.
>
> CVS:
> real 0m0.852s
> user 0m0.030s
> sys 0m0.130s
>
> SVN:
> real 0m4.840s
> user 0m1.680s
> sys 0m0.460s
I meant when committing when there are a lot of changed files, and you
don't specify them by name; or for that matter, updating from a highly
mixed working copy. But this is just by feel; it may be altogether
illusory :-).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: cvs/svn checkout/update timing info.
Posted by Michael Price <mp...@atl.lmco.com>.
Karl Fogel writes:
> Mix <mi...@acm.org> writes:
> > Kind of sad huh :)
> >
> > So then I went into each WC and performed an 'update' without having
> > changed a single thing. Here are the times:
> >
> > [...]
> >
> > Here are the times for a 'status' (and yes I know they aren't identical):
> >
> > [...]
>
> Yeah; as Branko says, this will improve when issue #749 is finished.
>
> Right now, the only place we win on speed is in commits. And "svn
> diff", of course :-).
I just did a 'echo "new file contents" > main.cc' in a cvs and svn WC and
then timed the commits. Same test setup as previous Mix email.
CVS:
real 0m0.852s
user 0m0.030s
sys 0m0.130s
SVN:
real 0m4.840s
user 0m1.680s
sys 0m0.460s
When the WC is on an NFS mount, I can't seem to find ANY operation where
svn is even close to cvs.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: cvs/svn checkout/update timing info.
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
Mix <mi...@acm.org> writes:
> Kind of sad huh :)
>
> So then I went into each WC and performed an 'update' without having
> changed a single thing. Here are the times:
>
> [...]
>
> Here are the times for a 'status' (and yes I know they aren't identical):
>
> [...]
Yeah; as Branko says, this will improve when issue #749 is finished.
Right now, the only place we win on speed is in commits. And "svn
diff", of course :-).
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org