1 |
On Jan 28, 2008 5:57 PM, Walter Dnes <waltdnes@××××××××.org> wrote: |
2 |
> On Mon, Jan 28, 2008 at 10:29:47PM +0900, Mike Mazur wrote |
3 |
> |
4 |
> > But the list of packages being recompiled have mostly to do with |
5 |
> > video, audio and transcoding. I understand it's the --newuse flag |
6 |
> > that's causing those, not the additional parameters in CFLAGS. Will |
7 |
> > the CFLAGS have benefits on other packages? Such as Firefox or maybe |
8 |
> > netscape-flash? For those I might want to do an emerge --emptytree |
9 |
> > world... |
10 |
> |
11 |
> If all the other stuff isn't being re-compiled, "-march=prescott" |
12 |
> probably includes them by default, so there's no point in re-building |
13 |
> your system. The CFLAGS were probably included by default. |
14 |
|
15 |
Portage *does not* look at CFLAGS in determining what to rebuild (even |
16 |
with -uDN) - portage only looks at USE flags and dependency |
17 |
upgrades/versions. Mike is correct in saying that, for packages to be |
18 |
recompiled with the new CFLAGS, he would have to recompile that |
19 |
package directly. emerge -e world is a good way to do this.\ |
20 |
|
21 |
-James |
22 |
|
23 |
|
24 |
> |
25 |
> If you need a speed boost in Firefox, there is some additional |
26 |
> tweaking that can be done. The pango library allows Firefox to |
27 |
> simultaneously render US English text (if that's your system locale) |
28 |
> *AND* Chinese, and other similar text. It slows down Firefox in the |
29 |
> process. If you're willing to give up on Asiatic text, you can cause |
30 |
> Firefox to not link against pango, by including the line... |
31 |
> |
32 |
> www-client/mozilla-firefox moznopango |
33 |
> |
34 |
> ...in /etc/portage/package.use It's your decision whether occasional |
35 |
> Asiatic scripts or a faster Firefox is worth more to you. |
36 |
|
37 |
Removing Pango will almost definitely increase the render/scroll speed |
38 |
of Firefox. |
39 |
|
40 |
However, from the symptoms that Mike is describing (system-wide |
41 |
momentary pauses, after which the system resumes normal |
42 |
responsiveness) sounds much more like a kernel-level issue - either |
43 |
I/O speed issues (check to make sure your hard drive is running at |
44 |
full speed and you have native controller supported compiled in to |
45 |
your kernel - also, what is your I/O scheduler set to by default?) or |
46 |
the Scheduler (process scheduler). What version of which kernel are |
47 |
you running? What does your .config look like? |
48 |
|
49 |
I had similar issues with momentarily frozen responsiveness on my |
50 |
laptop, until I upgraded to 2.6.23.x, which has the new CFS scheduler |
51 |
in it - seems to help responsiveness quite a bit. |
52 |
|
53 |
HTH |
54 |
|
55 |
-James |
56 |
-- |
57 |
gentoo-user@l.g.o mailing list |