This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: possible problems with gprof clocktick resolution when analyzing the program/library time ratio using gprof
- From: Daniel Jacobowitz <drow at false dot org>
- To: Mark Farnell <mark dot farnell at gmail dot com>
- Cc: binutils at sourceware dot org
- Date: Sun, 10 Jun 2007 21:29:07 -0400
- Subject: Re: possible problems with gprof clocktick resolution when analyzing the program/library time ratio using gprof
- References: <aaf549e0706080523u418a4ac3q962b9330ab0f394@mail.gmail.com>
On Sat, Jun 09, 2007 at 12:23:19AM +1200, Mark Farnell wrote:
> Then when I look at the output of gmon flat profile, I saw some
> function being called in the order of 100k times within 8s, that each
> call only take very very little time.
If that's the case, gprof is almost certainly the wrong tool for your
job. You may want to investigate oprofile or other
non-instrumentation profiling tools. Gprof tends to skew results for
frequently-called short-lived functions.
--
Daniel Jacobowitz
CodeSourcery