netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Eric Lemoine <eric.lemoine@gmail.com>
To: Stephen Hemminger <shemminger@osdl.org>
Cc: hadi@znyx.com, "David S. Miller" <davem@davemloft.net>,
	netdev <netdev@oss.sgi.com>
Subject: Re: [RFC] netif_rx: receive path optimization
Date: Thu, 31 Mar 2005 23:43:27 +0200	[thread overview]
Message-ID: <5cac192f05033113434742aeb4@mail.gmail.com> (raw)
In-Reply-To: <20050331131707.69f451ea@dxpl.pdx.osdl.net>

> > > Here is another alternative that seems better than the earlier posting. It uses
> > > a per device receive queue for non-NAPI devices.  The only issue is that then
> > > we lose the per-cpu queue's and that could impact the loopback device performance.
> > > If that is really an issue, then the per-cpu magic should be moved to the loopback
> > > device.
> > >
> >
> > The repurcassions of going from per-CPU-for-all-devices queue
> > (introduced by softnet) to per-device-for-all-CPUs maybe huge in my
> > opinion especially in SMP. A closer view of whats there now maybe
> > per-device-per-CPU backlog queue.
> 
> Any real hardware only has a single receive packet source (the interrupt routine),
> and the only collision would be in the case of interrupt migration.  So having
> per-device-per-CPU queue's would be overkill and more complex because
> the NAPI scheduling is per-netdevice rather than per-queue (though that
> could be fixed).
> 
> > I think performance will be impacted in all devices. imo, whatever needs
> > to go in needs to have some experimental data to back it
> 
> Experiment with what? Proving an absolute negative is impossible.
> I will test loopback and non-NAPI version of a couple of gigabit drivers
> to see.

Just a naive question : why at all trying to accelerate netif_rx?
Isn't NAPI the best choice for high performance rx anyway?


-- 
Eric

  parent reply	other threads:[~2005-03-31 21:43 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-03-30 21:28 [PATCH] netif_rx: receive path optimization Stephen Hemminger
2005-03-30 21:57 ` jamal
2005-03-30 22:08   ` jamal
2005-03-30 23:53   ` Stephen Hemminger
2005-03-31  3:16     ` jamal
2005-03-31 20:04 ` [RFC] " Stephen Hemminger
2005-03-31 21:10   ` Jamal Hadi Salim
2005-03-31 21:17     ` Stephen Hemminger
2005-03-31 21:25       ` Jamal Hadi Salim
2005-03-31 21:43       ` Eric Lemoine [this message]
2005-03-31 22:02         ` Stephen Hemminger
2005-03-31 21:24     ` Rick Jones
2005-03-31 21:38       ` jamal
2005-03-31 22:42         ` Rick Jones
2005-03-31 23:03           ` Nivedita Singhvi
2005-03-31 23:28             ` Rick Jones
2005-04-01  0:10               ` Stephen Hemminger
2005-04-01  0:42                 ` Rick Jones
2005-04-01  0:30               ` Nivedita Singhvi
2005-03-31 23:36           ` jamal
2005-04-01  0:07             ` Rick Jones
2005-04-01  1:17               ` jamal
2005-04-01 18:22                 ` Rick Jones
2005-04-01 16:40       ` Andi Kleen

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=5cac192f05033113434742aeb4@mail.gmail.com \
    --to=eric.lemoine@gmail.com \
    --cc=davem@davemloft.net \
    --cc=hadi@znyx.com \
    --cc=netdev@oss.sgi.com \
    --cc=shemminger@osdl.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;
as well as URLs for NNTP newsgroup(s).