PAGE_MAX_USE is just a constant which I will introduce
if I keep track of the number of times the page has
been used (this is just conceptual for now). You are
right page->age can never be greater than PAGE_AGE_MAX
(its a typo, the code should check for page->age ==
> > System Configuration
> > =====================
> > Single processor celeron system with 128 MB of
> RAM, running
> > Linux-2.4.7pre6 with Daniel's patch applied
> (running X windows at the
> > time of compilation, with GNOME).
> > time for creating clean bzImage *before* patch
> > ==============================================
> > real 28m40.492s
> > user 22m43.450s
> > sys 2m44.490s
> > time for creating clean bzImage *after* patch
> > =============================================
> > real 26m37.011s
> > user 21m56.350s
> > sys 2m28.060s
> Bash seems to have a built-in "time" command that
> isn't nearly as
> useful as the GNU version, which you'd find in
> /usr/bin/time. The GNU
> version tells us, among other things, how many swaps
> occured. Also,
> check the list for Marcelo's mm statistics patch.
> I'm not sure what
> the integration status is on that.
I will use this and get back to you, thanks for
letting me know this. I can probably ask you more
> > The system, seemed to respond faster (or I might
> be feeling so).
> > I am also planning to run some standard benchmark
> (I need to figure
> > out, which one, or you could guide me). If you
> like the idea, I will
> > post the benchmark results also to you (soon!).
> This patch is a
> > simple implementation of the idea, I could come
> out with a more
> > comprehensive solution if required.
> I imagine your system is swapping during your kernel
> build due to
> memory pressure created by gnome and X. If you show
> the swapping
> statistics from GNU time maybe we can suggest a more
> predictable way of
> creating a similar load. I always run my benchmark
> tests in text mode,
> by the way, just to try to eliminate some variables
> and get more
> consistent timing results.
> Did you repeat your timing measurements several
> times, and did you
> start each test with a clean reboot?
Yes I repeated the tests, but I guess and know there
is more to do.
Thanks for all the advice, will follow this up and
respond with more details
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to firstname.lastname@example.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/