public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* Re: oprofile results with lmbench
@ 2002-07-13 20:36 rwhron
  0 siblings, 0 replies; 2+ messages in thread
From: rwhron @ 2002-07-13 20:36 UTC (permalink / raw)
  To: akpm; +Cc: linux-kernel, lse-tech

> the numbers don't look right.  

> Perhaps you fed oprofpp the wrong vmlinux?

Thanks!  You are right. I just put a bold message
on the page that the results are invalid.  I'm leaving
the page up in case anyone else recommends changes.
I'll re-run the tests.

> btw, when using RETIRED_INSNS (or CPU_CLK_UNHALTED), you should
> boot the kernel with `idle=poll'.  Otherwise, idle time is
> unaccounted for in the profile and it gets really hard to interpret,
> and hard to compare runs.

I've added "idle=poll" to my lilo.conf append.

-- 
Rands Hron
http://home.earthlink.net/~rwhron/kernel/bigbox.html


^ permalink raw reply	[flat|nested] 2+ messages in thread
* oprofile results with lmbench
@ 2002-07-13 18:57 rwhron
  0 siblings, 0 replies; 2+ messages in thread
From: rwhron @ 2002-07-13 18:57 UTC (permalink / raw)
  To: lse-tech, linux-kernel

oprofile gives access to cpu performance counters on x86.

I ran oprofile with the individual lmbench tests to get
data from several of the counters.  This "hot kernel 
function" and counter info may be useful.  A web page with the
results from 5 of the performance counters for each
lmbench test is at:

http://home.earthlink.net/~rwhron/kernel/lmbench_oprofile.html

I'll add the rest of the counter data over the next few days.
Comments/caveats on how to make the information more useful are
welcome.

The target kernel was 2.4.19rc1aa1 on an Athlon 1333.

-- 
Randy Hron
http://home.earthlink.net/~rwhron/kernel/bigbox.html


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2002-07-13 20:34 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-07-13 20:36 oprofile results with lmbench rwhron
  -- strict thread matches above, loose matches on Subject: below --
2002-07-13 18:57 rwhron

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox