[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: pthreads inline functions break libgcc_s.so
|
From: |
Jeroen Dekkers |
|
Subject: |
Re: pthreads inline functions break libgcc_s.so |
|
Date: |
Tue, 27 Jul 2004 11:37:25 +0200 |
|
User-agent: |
Wanderlust/2.10.1 (Watching The Wheels) SEMI/1.14.6 (Maruoka) FLIM/1.14.6 (Marutamachi) APEL/10.6 Emacs/21.3.50 (i686-pc-linux-gnu) MULE/5.0 (SAKAKI) |
At 26 Jul 2004 16:07:27 -0700,
Thomas Bushnell, BSG wrote:
>
> Jeroen Dekkers <jeroen@dekkers.cx> writes:
>
> > I think we should just remove those inline functions, because it's
> > dubious whether they are a really faster and they do break
> > things. What do you think?
>
> How on earth is it dubious if they are really faster?
Inlining functions results in bigger code, which can lead to more
cache misses, especially on processors with small caches like the
celeron.
I've seen some article where the author compared the speed of Gentoo
versus Debian and Mandrake. He's compiling all programs on Gentoo with
-O3 (inlining a lot of functions) on his celeron and then compares the
speed with Debian and Mandrake and it is never the fastest.
Seems to be that the original article is removed from the web, but
google cache still has it:
http://216.239.41.104/search?q=cache:Ba3cFknoo4sJ:linuxfr.org/redirect/27426.html+%22many+people+recommend+-O3+for+this+kind+of+Celeron%22&hl=en
You can't be certain that inling is always faster, it depends on too
many factors.
Jeroen Dekkers