[Tfug] battery life, power management and windows vs linux

[Stephen Hooper]

On 3/23/07, t takahashi <gambarimasu at gmail.com> wrote:
> On 3/22/07, Stephen Hooper <stephen.hooper at gmail.com> wrote:
> > The only way I could suggest that writing atimes would slow down your
> > power consumption level to any radical extent would be if you weren't
> > writing to disk at all... in which case writing atimes seems to be a
> > little bit of a no brainer.
>
> one that the op did not mention having tried, stephen.

You mean the op didn't mention trying not to write to disk at all?

BTW, the use of my first name is quite a nice touch to any disparaging
email.  I read that as you are patient teacher trying to explain
something to an errant pupil.  Classy, and at the same time quite
catty.  Lovely!

>
> it is possible that the op is reading files, including binaries, that
> are cached, and the disk is spinning up just to write atime.  maybe
> even every single time the op runs /bin/echo, for example.  different
> file systems optimize in different ways, and it is possible that
> windows handles atimes differently from the op's linux fs, possibly
> affecting this type of thing.  there was a thread on the lkml not too
> long ago about this very topic, in fact.
>

Will I engage you in an argument about disk spin-ups professor?
Probably not... suffice it to say that both Windows, and Linux use a
"write back" cache.  It really doesn't matter what you think, but as
long as you are writing to disk you won't be seeing "extra" spin-ups.

If you were smart I guess you could even synchronize your writes to
your reads:  so maybe a disk spin-up is just a disk spin-up ;)

> even if the above is not the case, providing the hint to try enabling
> noatime on his fs could be worth the price of admission, regardless of
> how, or whether, his particular version of windows handles atime.  the
> latter isn't the point.  (please try not to be too literal about
> suggestions like this and assume they merely seek yes/no, any more
> than "what's cooking?" or "ever heard of politeness?" or "do you have
> the time?" need to be taken completely literally all of the time.)
> the former is the point.
>

I will try not to be too literal with suggestions like this in the
future:  the fact that I think it is a "persnickety" suggestion, and
one that won't really buy you much in battery life (even in the term
of minutes) is something I guess I should have not tried to be left
unsaid.

Anyways, I would turn on atime on my computer's root device, to prove
the point, but then I may be wrong, and would have to live with pigs
flying over a frozen hell :)

> did your post have a disparaging tone to it, stephen?  or did i
> misread it?  i'm hoping i just misread it.
>

Let's not misread this one.