Re: Removing xrdb for 10% startup win?

[Behdad Esfahbod <behdad cs toronto edu>]

On Mon, 29 Aug 2005, Lorenzo Colitti wrote:

> Alexander Larsson wrote:
> >> So, as Owen says, most of the time is spent just reading the cc1
> >> binary into memory. I have no idea why it's reading around all over
> >> the place instead of performing one big sequential read of the whole
> >> file though.
> >
> > My guess is that it just maps the whole executable on start, and then
> > demand loads pages as they are needed (i.e. on page faults).
>
> Yes, I suppose you're right! Thanks for enlightening me.
>
> However, that's not the most efficient way to load *anything* if you
> know you're going to need the whole file. I see this happening with
> shared libraries a lot.
>
> But surely someone has thought of this before?

That's in fact my Summer of Code project :-).
http://preload.sf.net/


> Cheers,
> Lorenzo

--behdad
http://behdad.org/