| … | |
… | |
| 12 | <author title="Editor"><mail link="peesh@gentoo.org">Jorge Paulo</mail></author> |
12 | <author title="Editor"><mail link="peesh@gentoo.org">Jorge Paulo</mail></author> |
| 13 | <author title="Editor"><mail link="swift@gentoo.org">Sven Vermeulen</mail></author> |
13 | <author title="Editor"><mail link="swift@gentoo.org">Sven Vermeulen</mail></author> |
| 14 | <abstract>This FAQ is a collection of questions and answers collected from the gentoo-dev mailing list and from IRC -- if you have any questions (or answers!) to add, please contact either an author or a member of the documentation team.</abstract> |
14 | <abstract>This FAQ is a collection of questions and answers collected from the gentoo-dev mailing list and from IRC -- if you have any questions (or answers!) to add, please contact either an author or a member of the documentation team.</abstract> |
| 15 | |
15 | |
| 16 | <version>1.1.4</version> |
16 | <version>1.1.4</version> |
| 17 | <date>5 May 2003</date> |
17 | <date>6 May 2003</date> |
| 18 | |
18 | |
| 19 | <chapter> |
19 | <chapter> |
| 20 | <title>Getting Started</title> |
20 | <title>Getting Started</title> |
| 21 | |
21 | |
| 22 | <section> |
22 | <section> |
| … | |
… | |
| 99 | -fomit-frame-pointer" optimizations. What gives?</title> |
99 | -fomit-frame-pointer" optimizations. What gives?</title> |
| 100 | <body> |
100 | <body> |
| 101 | <p>When you use any optimizations beyond <c>-O3</c>, you're really taking the risk of |
101 | <p>When you use any optimizations beyond <c>-O3</c>, you're really taking the risk of |
| 102 | having broken packages. |
102 | having broken packages. |
| 103 | Very aggressive optimizations sometimes cause the compiler to streamline the assembly code |
103 | Very aggressive optimizations sometimes cause the compiler to streamline the assembly code |
| 104 | to the point where it doesn't quite do the same thing anymore. <e>Loc-Dog</e> (on IRC) |
104 | to the point where it doesn't quite do the same thing anymore. A possible setting based on <e>Loc-Dog</e> (on IRC)'s CFLAGS |
| 105 | uses <c>-O3 -mcpu=i686 -march=i686 -fforce-addr -fomit-frame-pointer -funroll-loops |
105 | is <c>-O3 -mcpu=i686 -march=i686 -fforce-addr -fomit-frame-pointer -funroll-loops |
| 106 | -frerun-cse-after-loop -frerun-loop-opt -malign-functions=4</c>, which is about |
106 | -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4</c>, which is about |
| 107 | as much as I'd want to push global optimization settings. Beyond this, it's best to use |
107 | as much as I'd want to push global optimization settings. Beyond this, it's best to use |
| 108 | ultra-high optimizations only with specific packages where you really need that extra 2%, |
108 | ultra-high optimizations only with specific packages where you really need that extra 2%, |
| 109 | (eg graphics and various multimedia programs), and where you can easily test the package |
109 | (eg graphics and various multimedia programs), and where you can easily test the package |
| 110 | to ensure that it hasn't been optimized into oblivion.</p> |
110 | to ensure that it hasn't been optimized into oblivion.</p> |
| 111 | <p>Please try first to compile with CFLAGS <c>-march= -O2</c> before reporting a bug</p> |
111 | <p>Please try first to compile with CFLAGS <c>-march= -O2</c> before reporting a bug</p> |
Parent Directory
| 
Revision Log
| 
Patch