1 |
Koby Boy wrote: |
2 |
> On Tue, 2003-05-06 at 03:24, eNTi wrote: |
3 |
>>CFLAGS="-mcpu=athlon-xp -O3 -pipe -m3dnow -msse -mmmx -Wall -fomit-frame-pointer" / CHOST="i686-pc-linux-gnu" |
4 |
> |
5 |
> I've got an Athon-XP 2400 (2 GHz) but I've over clocked it to the |
6 |
> equivalent of an Athlon-XP 2600 (2.2GHz). My CFLAGS look like this: |
7 |
> |
8 |
> CFLAGS="-mcpu=athlon-xp -march=athlon-xp -O3 -fforce-addr |
9 |
> -fomit-frame-pointer -funroll-loops -frerun-cse-after-loop |
10 |
> -frerun-loop-opt -falign-functions=4 -foptimize-sibling-calls |
11 |
> -fexpensive-optimizations -pipe -m3dnow -mmmx -msse -mfpmath=sse,387" |
12 |
|
13 |
-march=athlon-xp implies: |
14 |
-mcpu=athlon-xp |
15 |
|
16 |
-march=athlon-xp implies: |
17 |
-m3dnow |
18 |
-mmmx |
19 |
-msse |
20 |
|
21 |
-O3 implies -O2 which implies: |
22 |
-frerun-loop-opt |
23 |
-frerun-cse-after-loop |
24 |
-frerun-loop-opt |
25 |
-falign-functions |
26 |
-fexpensive-optimizations |
27 |
|
28 |
-mfpmath=sse,387 is still unstable.. asking for trouble! |
29 |
|
30 |
Lastly, many people have expressed that -funroll-loops slows down more |
31 |
code than it speeds up. |
32 |
|
33 |
I'd be surprised if you see much improvement over '-O3 -pipe', in 99% of |
34 |
apps, not worth it's time in debugging broken code :/ |
35 |
|
36 |
man gcc *g* |
37 |
|
38 |
MAL |
39 |
|
40 |
|
41 |
-- |
42 |
gentoo-performance@g.o mailing list |