netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andi Kleen <ak@muc.de>
To: "David S. Miller" <davem@davemloft.net>
Cc: jgarzik@pobox.com, netdev@oss.sgi.com
Subject: Re: Intel and TOE in the news
Date: Sat, 19 Feb 2005 21:27:31 +0100	[thread overview]
Message-ID: <m1psywb8i4.fsf@muc.de> (raw)
In-Reply-To: <20050219114624.373af63f.davem@davemloft.net> (David S. Miller's message of "Sat, 19 Feb 2005 11:46:24 -0800")

"David S. Miller" <davem@davemloft.net> writes:

> Read what they are doing.  The idea is not to have all of this network
> protocol logic off-cpu, the idea is to "reduce some of the time
> a processor typically spends waiting for memory to feed back information"
>
> Think about what part of the network I/O equation that is working on.
> It's not protocol offload, that's for sure.

<speculating freely>

It would be nice if the NIC could asynchronously trigger prefetches in
the CPU. Currently a lot of the packet processing cost goes
to waiting for read cache misses.

E.g. 

- NIC receives packet. 
- Tells target CPU to prefetch RX descriptor and headers. 
- CPU later looks at them and doesn't have to wait a for a cache miss.

Drawback is that you would need to tell the NIC in advance
on which CPU you want to process the packet, but with Linux
IRQ affinity that's easy to figure out. 

-Andi

  reply	other threads:[~2005-02-19 20:27 UTC|newest]

Thread overview: 74+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-02-19  3:44 Intel and TOE in the news Jeff Garzik
2005-02-19  4:10 ` Lennert Buytenhek
2005-02-19 19:46   ` David S. Miller
2005-02-19 20:27     ` Andi Kleen [this message]
2005-02-19 20:32       ` Lennert Buytenhek
2005-02-20 16:46       ` Eugene Surovegin
2005-02-21 14:01         ` jamal
2005-02-20 19:45       ` rick jones
2005-02-20 21:20         ` Michael Richardson
2005-02-20 21:29         ` Andi Kleen
2005-02-20 22:43           ` Leonid Grossman
2005-02-20 23:07             ` Andi Kleen
2005-02-21  1:57               ` Alex Aizman
2005-02-21  2:37                 ` Jeff Garzik
2005-02-21 19:34                   ` Alex Aizman
2005-02-21 20:34                     ` Jeff Garzik
2005-02-22  0:50                       ` Alex Aizman
2005-02-21 11:37                 ` Andi Kleen
2005-02-21  3:31               ` Leonid Grossman
2005-02-21 11:50                 ` Andi Kleen
2005-02-21 13:28                   ` Thomas Graf
2005-02-21 14:03                     ` jamal
2005-02-21 14:17                       ` Thomas Graf
2005-02-21 14:31                         ` jamal
2005-02-21 15:34                           ` Thomas Graf
2005-02-21 15:48                             ` jamal
2005-02-21 16:40                               ` Thomas Graf
2005-02-21 17:03                                 ` jamal
2005-02-21 20:12                                   ` patrick mcmanus
2005-02-21 21:12                                     ` jamal
2005-03-06 11:21                                       ` Harald Welte
2005-02-21 21:41                                   ` Thomas Graf
2005-02-21 15:38                           ` Robert Olsson
2005-02-21 15:50                             ` jamal
2005-02-21 13:44             ` jamal
2005-02-21 16:52               ` Leonid Grossman
2005-02-21 17:11                 ` jamal
2005-02-21 18:02                   ` Leonid Grossman
2005-02-22 18:02                     ` Stephen Hemminger
2005-02-22 18:07                       ` Andi Kleen
2005-02-22 20:51                         ` Leonid Grossman
2005-02-22 21:20                           ` Rick Jones
2005-02-22 21:30                             ` Leonid Grossman
2005-02-22 21:42                               ` Rick Jones
2005-02-22 22:10                                 ` Leonid Grossman
2005-02-22 21:43                           ` Andi Kleen
2005-02-22 22:17                             ` Leonid Grossman
2005-02-22 22:42                               ` Andi Kleen
2005-02-22 22:51                                 ` Leonid Grossman
2005-03-14 20:22                   ` [ANNOUNCE] Experimental Driver for Neterion/S2io 10GbE Adapters Alex Aizman
2005-03-14 20:38                     ` David S. Miller
2005-03-14 20:53                       ` Leonid Grossman
2005-03-14 23:27                         ` Andi Kleen
2005-03-14 23:45                           ` Jeff Garzik
2005-03-15  0:32                             ` Leonid Grossman
2005-03-15  1:07                           ` Alex Aizman
2005-03-15  1:29                             ` Rick Jones
2005-03-15  2:28                               ` Leonid Grossman
2005-03-15 15:07                             ` Leonid Grossman
2005-03-15 15:55                               ` Leonid Grossman
2005-03-19 20:15                                 ` Andi Kleen
2005-03-19 22:19                                   ` Leonid Grossman
2005-03-20 13:40                                     ` jamal
2005-03-20 20:13                                       ` Leonid Grossman
2005-03-15  5:14                     ` Scott Feldman
2005-03-15  5:59                       ` Matt Mackall
2005-03-15  6:02                       ` Leonid Grossman
2005-02-22 17:27                 ` Intel and TOE in the news Andi Kleen
2005-02-19 20:29     ` Lennert Buytenhek
2005-03-02 13:48   ` Lennert Buytenhek
2005-03-02 17:34     ` Leonid Grossman
2005-02-21 13:59 ` P
2005-02-21 14:10   ` jamal
2005-02-21 22:44 ` 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=m1psywb8i4.fsf@muc.de \
    --to=ak@muc.de \
    --cc=davem@davemloft.net \
    --cc=jgarzik@pobox.com \
    --cc=netdev@oss.sgi.com \
    /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).