[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Parallel Booting



On Fri, Jul 06, 2007 at 05:56:35PM -0400, Colin Walters wrote:
> Ok, I just googled and found a really good thread:
> http://kerneltrap.org/node/2157
> 
> It's unclear to me though how much of this is actually running now.

Quite a bit in terms of the disk layer. We readahead and writebehind,
we preallocate data so that trickling writes don't end up smeared around the
disk

> 1) Our precaching isn't as good as their implementation
> 2) MacOS X is fundamentally different
> 
> I think it's more likely to be 1.  I could be wrong though.

Try running FC7 off a Gigabyte i-RAM (you'll need the approved RAM and
a recent firmware for it to be reliable). You can then seperate the
disk and non-disk load

> I believe the icon cache is currently
> mmap'd: /usr/share/icons/hicolor/icon-theme.cache
> Whether it's your actual code or not I don't know.

Hooray, and I hope mapped shared across all users.

> > Unfortunately most
> > of the mess *is* userspace 
> 
> Right.  That doesn't change the fact that some kernel work seems very
> likely to speed things up quite a bit.

Yes. There are things we can do such as better predicting what I/O is
coming and what data not to write yet (and when to defer writing for
reading). There are things we can't do like make the old gconf one key
per file implementation not suck.

> > complete denial. Just profile the number of file opens of different files
> > done in a gnome startup and when you've finished laughing you can weep.
> 
> Work continues on improving userspace:
> http://primates.ximian.com/%7Efederico/news-2007-06.html#26

Which is good. 

One area that probably could do with more work (although openoffice has
forced many improvements ;)) is the linker and ELF loader side, both from
the user space point of view for performance, and the kernel point of view
to reduce page fault counts and between the two of them to load data in a
sane order when doing demand paging (actually for large systems I question
these days how useful demand paging is for most apps)


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]