SoC boot project weekly summary week #3/14

Petter Reinholdtsen pere at
Sat Jun 17 22:43:43 UTC 2006

[Bas Zoetekouw]
> I think on typical desktop systems (at least on mine), lots of time
> is used in setting up lvm and mounting (large) disks.  Any way that
> could be measured too? (e.g., compare boot speed with a large
> reiserfs, ext2/3, xfs or vfat partition)

Well, I guess you can run bootchart on such systems and compare their
boot time?

> I'm not sure the DE starting should be counted at all.  The main issue
> at hand is the boot itself, right?  I think Gnome and KDE have their own
> benchmarks and ways to increase log in times...

Well, for a users point of view, the boot time is from the machine is
turned on and until it is ready to do real work.  So from that
perspective, the desktop environment should be measured as well.
There are claims that readahead have most impact on the start time for
the desktop, so I believe it is very relevant when measuring the boot

> Agreed.  I didn't know that providing the dependency info was already
> possible in Debian.  Is it documented somewhere (policy, dev reference)?  

It is documented in <URL:>.  I
guess it should make it into the developers reference.

Petter Reinholdtsen

More information about the initscripts-ng-devel mailing list