mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Magnus Damm" <magnus.damm@gmail.com>
To: "Jesper Juhl" <jesper.juhl@gmail.com>
Cc: "Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>
Subject: Re: SMP and 101% cpu max?
Date: Wed, 8 Mar 2006 10:53:02 +0900	[thread overview]
Message-ID: <aec7e5c30603071753x6c6b413dj135d6589003d0229@mail.gmail.com> (raw)
In-Reply-To: <9a8748490603070507h48e2fe02qbf9da7956e794161@mail.gmail.com>

On 3/7/06, Jesper Juhl <jesper.juhl@gmail.com> wrote:
> On 3/7/06, Magnus Damm <magnus.damm@gmail.com> wrote:
> > Hi guys,
> >
> > I'm doing some memory related hacking, and part of that is testing the
> > current behaviour of the Linux VM. This testing involves running some
> > simple tests (building the linux kernel is one of the tests) while
> > varying the amount of RAM available to the kernel.
> >
> > I've tested memory configurations of 64MB, 128MB and 256MB on a Dual
> > PIII machine. The tested kernel is 2.6.16-rc5, and user space is based
> > on debian-3.1. I run 5 tests per memory configuration, and the machine
> > is rebooted between each test.
> >
> > Problem:
> >
> > With 128MB and 256MB configurations, a majority of the tests never
> > make it over 101% CPU usage when I run "make -j 2 bzImage", building a
> > allnoconfig kernel. With 64MB memory, everything seems to be working
> > as expected. Also, running "make bzImage" works as expected too.
> >
> Hmm, I wonder if it's related to the problem I reported here :
> http://lkml.org/lkml/2006/2/28/219
> Where I need to run make -j 5 or higher to load both cores of my Athlon X2.

Yeah, the problem looks kind of related. But I doubt it is the CPU
scheduler only that plays tricks on us, because in the my 64MB case
things seem to work ok at all times. But maybe it's just random.

I've spent this morning testing 2.6.15, and that version seems to work
ok. And just to double check I rebuilt and retested 2.6.16-rc5. This
round gives me the similar results - now it fails to go above 101% CPU
load in one case.

Results for "make -j 2 bzImage", 2.6.15:

time: 123.527 220.11 20.27, cpu: 195, ram: 126896, swap: 125944
time: 123.823 219.48 20.57, cpu: 194, ram: 126896, swap: 125944
time: 123.867 221.1 19.35, cpu: 194, ram: 126896, swap: 125944
time: 124.6 220.49 20.25, cpu: 193, ram: 126896, swap: 125944
time: 124.96 222.14 20.18, cpu: 194, ram: 126896, swap: 125944

Results for "make -j 2 bzImage", 2.6.16-rc5:

time: 125.045 220.02 18.71, cpu: 191, ram: 126876, swap: 125944
time: 125.052 221.93 18.35, cpu: 192, ram: 126876, swap: 125944
time: 125.598 220.16 19.34, cpu: 191, ram: 126876, swap: 125944
time: 125.7 220.07 19.32, cpu: 190, ram: 126876, swap: 125944
time: 229.316 212.86 17.72, cpu: 101, ram: 126876, swap: 125944

/ magnus

  parent reply	other threads:[~2006-03-08  1:53 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-03-07 12:34 Magnus Damm
2006-03-07 13:07 ` Jesper Juhl
2006-03-07 13:20   ` [PATCH] IRQ: prevent enabling of previously disabled interrupt Bryan Holty
2006-03-07 23:58     ` Andrew Morton
2006-03-08  1:53   ` Magnus Damm [this message]
2006-03-08  2:50   ` FWIW: Re: SMP and 101% cpu max? Linda Walsh
2006-03-08  3:04     ` Magnus Damm

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aec7e5c30603071753x6c6b413dj135d6589003d0229@mail.gmail.com \
    --to=magnus.damm@gmail.com \
    --cc=jesper.juhl@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®