1 |
Hi, |
2 |
|
3 |
On Mon, 20 Nov 2006 23:42:06 -0600 "Boyd Stephen Smith Jr." |
4 |
<bss03@××××××××××.net> wrote: |
5 |
|
6 |
> On Monday 20 November 2006 15:13, Daniel Iliev <danny@××××××××.com> |
7 |
> wrote about 'Re: [gentoo-user] mplayer compilation': |
8 |
> > I don't think there is such an option for gcc "-O4". AFAIK the |
9 |
> > shorthand optimization options are -O, -O0, -O1, -O2, -O3 and -Os" |
10 |
> |
11 |
> In addition to -Os, -O will accept any non-negative integer as an |
12 |
> argument. Currently, anything above 3 does nothing extra. From what |
13 |
> I understand, there have been undocumented levels up to 6, but they |
14 |
> have always been considered unstable (which is why they never |
15 |
> appeared in documentation). |
16 |
|
17 |
Hm, I don't know if they we're considerend unstable, but I think they |
18 |
were too much of a moving target to actually document them. The -O |
19 |
parameters are passed to gcc's code modules. |
20 |
|
21 |
> The quickest way to recognize a Gentoo ricer is someone that |
22 |
> has -O4, -O6, -O99 or something else equally stupid in their CFLAGS. |
23 |
> -O3 is defensible, as it is at least *supposed to be* stable; |
24 |
> anything higher is just stupid. |
25 |
|
26 |
OK, I agree that this might be an indication that the person just |
27 |
c&p'ed a part of a wannabee-howto from the forums... |
28 |
|
29 |
In this case, however, it's the default setting (at least on my |
30 |
machine) that mplayer choses when it does its own recognizing of the |
31 |
best settings (and ignores Gentoo's CFLAGS). And I've yet to see it |
32 |
break... Myself, I would never care about spending my CPU power beyond |
33 |
anything than -O2, where -O1 is mostly absolutely sufficient. I've seen |
34 |
broken compile runs for -Os and -O3, but -O2 is the safe thing. |
35 |
|
36 |
-hwh |
37 |
-- |
38 |
gentoo-user@g.o mailing list |