Ethernet Bridge development
 help / color / mirror / Atom feed
From: Brad Dameron <brad.dameron@clearwire.com>
To: bridge@lists.linux-foundation.org
Subject: RE: [Bridge] Lowering CPU utilisation
Date: Tue, 06 Nov 2007 14:19:51 -0800	[thread overview]
Message-ID: <1194387591.7458.0.camel@linux-2n2q.site> (raw)
In-Reply-To: <1194387232.7283.5.camel@linux-2n2q.site>

On Tue, 2007-11-06 at 14:13 -0800, Brad Dameron wrote:
> On Wed, 2007-11-07 at 08:51 +1100, Leigh Sharpe wrote:
> > I'm using e1000 cards.
> > According to oprofile, around 50% of the time is spent in the e1000
> > driver. 
> > Given that my load average is still under 0.5, I suspect I've got
> > plenty of room for the driver to handle more traffic before I need to
> > worry too much.
> >  
> > 
> > 
> > ______________________________________________________________________
> > From: Vaidas M [mailto:admin@vdx.lt] 
> > Sent: Tuesday, 6 November 2007 9:32 PM
> > To: Leigh Sharpe
> > Subject: RE: [Bridge] Lowering CPU utilisation
> > 
> > 
> > 
> > What kind of NIC you are using?
> > 
> > I suggest you to use cards with NAPI featured in driver.
> > 
> >  
> > 
> >                         _      _     _    _   _  _ _ __ ___ ____ _____
> > 
> >                                                             Vaidas M. 
> > 
> > 
> >  
> > 
> > From: bridge-bounces@lists.linux-foundation.org
> > [mailto:bridge-bounces@lists.linux-foundation.org] On Behalf Of Leigh
> > Sharpe
> > Sent: Thursday, November 01, 2007 12:00 AM
> > To: bridge@lists.linux-foundation.org
> > Subject: [Bridge] Lowering CPU utilisation
> > 
> > 
> >  
> > 
> > Hi All, 
> > 
> > 
> >  I have a box with 12 bridges on it, each of which has 2 ports.
> > 
> > 
> > When the traffic level on the box reaches around 15-20Mbps on each
> > bridge, my CPU utilisation gets to about 90-95%.
> > 
> > 
> > I would like to avoid working the CPU that hard. Is there anything
> > which can be done to get the CPU utlilisation down a bit? I have plans
> > for an even bigger bridge, but these results indicate that it may not
> > scale quite that well.
> > 
> > 
> >  
> > 
> > 
> > Regards,
> >              Leigh
> > 
> 
> 
> I assume you are using Linux. Try setting these in your /etc/sysctl.conf
> file.
> 
> # increase TCP max buffer size
> net.core.rmem_max = 8388608
> net.core.wmem_max = 8388608
> # increase Linux autotuning TCP buffer limits
> # min, default, and max number of bytes to use
> net.ipv4.tcp_mem = 8388608 8388608 8388608
> net.ipv4.tcp_rmem = 4096 1048576 8388608
> net.ipv4.tcp_wmem = 4096 65535 8388608
> net.core.optmem_max = 40960
> 
> 
> Then load them with sysctl -p command.
> 
> Hope this helps. 

Oops. Too many lists. Of course your using Linux.


Brad Dameron
Senior Systems Engineer
Clearw're



  reply	other threads:[~2007-11-06 22:19 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-11-06 21:51 [Bridge] Lowering CPU utilisation Leigh Sharpe
2007-11-06 22:13 ` Brad Dameron
2007-11-06 22:19   ` Brad Dameron [this message]
2007-11-06 22:18 ` Stephen Hemminger
  -- strict thread matches above, loose matches on Subject: below --
2007-11-07  2:20 Leigh Sharpe
2007-10-31 22:00 Leigh Sharpe
2007-10-31 22:08 ` Stephen Hemminger

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=1194387591.7458.0.camel@linux-2n2q.site \
    --to=brad.dameron@clearwire.com \
    --cc=bridge@lists.linux-foundation.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