Changes in / [d2eb97b:3eb188e]
- Files:
-
- 3 edited
Legend:
- Unmodified
- Added
- Removed
-
chapter04/aboutsbus.xml
rd2eb97b r3eb188e 27 27 28 28 <para>For example, consider a package whose compilation time is 4.5 29 SBUs. This means that if your system took 4minutes to compile and29 SBUs. This means that if your system took 10 minutes to compile and 30 30 install the first pass of binutils, it will take 31 <emphasis>approximately</emphasis> 18minutes to build the example package.31 <emphasis>approximately</emphasis> 45 minutes to build the example package. 32 32 Fortunately, most build times are shorter than one SBU.</para> 33 33 … … 37 37 numbers can vary by as much as dozens of minutes in some cases.</para> 38 38 39 <para>On some newer systems, the motherboard is capable of contolling 40 the system clock speed. This can be controlled with a command such as 41 <command>powerprofilesctl</command>. This is not available in LFS, but 42 may be available on the host distro. After LFS is complete, it can be 43 added to a system with the procedures at the 44 <ulink url='&blfs-book;sysutils/power-profiles-daemon.html'> 45 BLFS power-profiles-daemon</ulink> page. 39 <para>Before measuring the build time of any package (no matter Binutils 40 pass 1 or a package of which the SBU is being measured), make sure a 41 system power profile suitable to make the system running with the 42 maximum performance (and the maximum power consumption) is selected. Or 43 the measured SBU value may be severly inaccurate because the system may be 44 operated differently building Binutils pass 1 and the other package. 45 Be aware that a significant inaccuracy can still show up even if the same 46 profile (except one maximizing the performance) is used for both packages: 47 the system may respond slower for <quote>saving the power</quote> building 48 Binutils pass 1, because the system load seems only about 25% of the load 49 building the other package (with <parameter>-j4</parameter>).</para> 46 50 47 Before measuring the build time of any package it is advisable to use a 48 system power profile set for maximum performance (and maximum power 49 consumption). 51 <para>On most distros the power profile can be managed with either 52 <command>power-profiles-daemon</command> or <command>tuned</command>. 53 If the distro runs <command>power-profiles-daemon</command>, issue the 54 <command>powerprofilesctl set performance</command> command to select 55 the <literal>performance</literal> profile. If the distro runs 56 <command>tuned</command>, issue the <command>tuned-adm profile 57 throughput-performance</command> command to select the 58 <literal>throughput-performance</literal> profile.</para> 50 59 51 Otherwise the measured SBU value may be inaccurate because the 52 system may react differently when building <xref linkend='ch-tools-binutils-pass1'/> 53 or other packages. 54 55 Be aware that a significant inaccuracy can still show up even if the same 56 profile is used for both packages because the system may respond slower if 57 the system is idle when starting the build procedure. Setting the power 58 profile to <quote>performance</quote> will minimize this problem. And 59 obviously doing so will also make the system build LFS faster.</para> 60 61 <para>If <command>powerprofilesctl</command> is available, issue the 62 <command>powerprofilesctl set performance</command> command to select 63 the <literal>performance</literal> profile. Some distros provides the 64 <command>tuned-adm</command> command for managing the profiles instead of 65 <command>powerprofilesctl</command>, on these distros issue the 66 <command>tuned-adm profile throughput-performance</command> command to 67 select the <literal>throughput-performance</literal> profile.</para> 60 <para>Even if you are not measuring the SBU values, it's still better to 61 select the power profile for maximum performance before building LFS, as 62 doing so can (obviously) make the system faster to build LFS 63 packages.</para> 68 64 69 65 <note> … … 75 71 single processor build to properly analyze the error messages.</para> 76 72 77 <para>The times presented here for all packages 78 (except <xref linkend='ch-tools-binutils-pass1'/> which is based on one core) 79 are based upon using four cores (-j4). The 73 <para>The times presented here are based upon using four cores (-j4). The 80 74 times in Chapter 8 also include the time to run the regression tests for 81 75 the package unless specified otherwise.</para> -
chapter05/binutils-pass1.xml
rd2eb97b r3eb188e 6 6 ]> 7 7 8 <sect1 id="ch-tools-binutils-pass1" role="wrap" xreflabel="binutils-pass1">8 <sect1 id="ch-tools-binutils-pass1" role="wrap"> 9 9 <?dbhtml filename="binutils-pass1.html"?> 10 10 -
chapter08/gcc.xml
rd2eb97b r3eb188e 162 162 </important> 163 163 164 <para>GCC may need more stack space compiling some extremely complex 165 code patterns. As a precaution for the host distros with a tight stack 166 limit, explicitly set the stack size hard limit to infinite. 167 On most host distros (and the final LFS system) the hard limit is 168 infinite by default, but there is no harm done by setting it explicitly. 169 It's not necessary to change the stack size soft limit because GCC will 170 automatically set it to an appropriate value, as long as the value does 171 not exceed the hard limit:</para> 172 173 <screen><userinput remap="test">ulimit -s -H unlimited</userinput></screen> 164 <para>One set of tests in the GCC test suite is known to exhaust the default 165 stack, so increase the stack size prior to running the tests:</para> 166 167 <screen><userinput remap="test">ulimit -s 32768</userinput></screen> 174 168 175 169 <para>Now remove/fix several known test failures:</para>
Note:
See TracChangeset
for help on using the changeset viewer.