linux-mips
[Top] [All Lists]

Re: [PATCH 1/7] signals: reduce {setup,restore}_sigcontext sizes

To: Franck Bui-Huu <vagabon.xyz@gmail.com>
Subject: Re: [PATCH 1/7] signals: reduce {setup,restore}_sigcontext sizes
From: Ralf Baechle <ralf@linux-mips.org>
Date: Tue, 23 Jan 2007 16:36:46 +0000
Cc: linux-mips@linux-mips.org, Franck Bui-Huu <fbuihuu@gmail.com>
In-reply-to: <cda58cb80701230826i3cba9164jf20678f9efd1a7ba@mail.gmail.com>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <1169561903878-git-send-email-fbuihuu@gmail.com> <11695619031540-git-send-email-fbuihuu@gmail.com> <20070123143814.GE18083@linux-mips.org> <cda58cb80701230826i3cba9164jf20678f9efd1a7ba@mail.gmail.com>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.4.2.2i
On Tue, Jan 23, 2007 at 05:26:21PM +0100, Franck Bui-Huu wrote:

> No, I haven't. Since the size code has been reduced by a factor 2, I
> would think that signal code can better fit in instruction cache
> lines. For example, the loop is made up by 11 instructions (I don't
> know why gcc makes it so big though) which fits into 3 cache lines in
> my cases. Where as the old code generated 246 instructions for the
> same job, which should cause many more cache misses.
> 
> Do you have any pointers on benchmarks I could run ?

For stuff like this microbenchmarks like lmbench are best suited.  Lmbench
recently moved to http://sourceforge.net/projects/lmbench.

  Ralf

<Prev in Thread] Current Thread [Next in Thread>