diff options
Diffstat (limited to 'chapter02/aboutsbus.xml')
| -rw-r--r-- | chapter02/aboutsbus.xml | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/chapter02/aboutsbus.xml b/chapter02/aboutsbus.xml new file mode 100644 index 000000000..d78e6b719 --- /dev/null +++ b/chapter02/aboutsbus.xml @@ -0,0 +1,30 @@ +<sect1 id="ch02-aboutsbus"> +<title>About SBUs</title> +<?dbhtml filename="aboutsbus.html" dir="chapter02"?> + +<para>SBUs are <emphasis>Static Bash Units</emphasis> and they are our way +of identifying how long a package takes to compile. Why don't we use normal +times like anybody else?</para> + +<para>The biggest problem is that times cannot be acurate, not even a +little bit. So many people install LFS on so many different systems, the +times it takes to compile something varies too much. One package may take +20 minutes on one system, but that same packages may take 3 days on another +(this is not an exaggeration). So instead we've come up with a +<emphasis>Static Bash Unit</emphasis> or <emphasis>SBU</emphasis>.</para> + +<para>It works like this: the very first package you compile in this book +is Bash in chapter 5 and it'll be statically linked. The time it takes to +compile this package will be the basis and called the SBU. All other +compile times are relative to the time it takes to install Bash. For +example, GCC-3.2 takes about 9.5 SBUs and it's proven that this number if +fairly consistent among a lot of different systems. So multiply 9.5 by the +number of seconds it takes for Bash to install (the SBU value) and you get +a close approximation of how long GCC will take on your system.</para> + +<para>Note: SBUs don't work on SMP machines. We've seen that SBUs don't +work well on SMP based machines. So all bets are off if you're lucky enough +to have an SMP setup.</para> + +</sect1> + |