Re: Starting fresh on my Toshiba A105-S2101 and want a custom kernel
Erik Sundin <sudden@vilse.eu> writes:
> Den Wednesday
> 08 April 2009 03.44.30 skrev Daniel Pittman:
>> Preston Boyington <preston.lists@gmail.com> writes:
>> > After happily using my laptop for the last couple years I've decided to
>> > streamline it a good bit. My hard disk that contains the existing Debian
>> > Sid install has been replaced with a clean drive.
>> > I believe I could get some speed with a custom kernel
>> That is extremely unlikely. Is there any particular reason you believe that
>> a custom kernel would be faster?
[...]
> I've been compiling my own kernels since 2.6.18 and though I really
> don't get much of a change in performance boot time can be shortened
> quite a bit (in my experience) by a 'custom' kernel.
The investigation done by Ubuntu supports that: they found that building
in a set of core drivers did, indeed, reduce boot time.[1]
The OP didn't mention anything about boot time, though, and the most
commonly cited reason for a custom kernel (in my experience) is "speed",
where the poster assumes that a generic kernel must, necessarily, be
slower than a custom one.
> I started doing it mostly to learn how it's done and to try out some
> "fancy" stuff with the intel chips that's been in most of my
> laptops. Since I don't care very much 'bout that whole 'stability'
> nonsense I pull the git tree once every to weeks and then compile it
> (spare cycles... whith the amount of power in modern cpus, why not?).
> I might add that running a 'slightly' newer kernel then what's
> available in debian actually sometimes adds some 'performance' (and
> alot of excitement...)
To each their own, I suppose. ;)
Regards,
Daniel
Footnotes:
[1] I don't quite understand /why/ this is, but hey, presumably
something is inefficient in the modprobe dynamic linking process
that slows things down.
Reply to: