From owner-freebsd-hackers  Thu Jun 19 01:22:54 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id BAA15760
          for hackers-outgoing; Thu, 19 Jun 1997 01:22:54 -0700 (PDT)
Received: from misery.sdf.com (misery.sdf.com [204.244.210.193])
          by hub.freebsd.org (8.8.5/8.8.5) with SMTP id BAA15754
          for <hackers@freebsd.org>; Thu, 19 Jun 1997 01:22:49 -0700 (PDT)
Received: from tom by misery.sdf.com with smtp (Exim 1.62 #1)
	id 0wecRg-0003Hn-00; Thu, 19 Jun 1997 01:19:56 -0700
Date: Thu, 19 Jun 1997 01:19:55 -0700 (PDT)
From: Tom Samplonius <tom@sdf.com>
To: sthaug@nethelp.no
cc: ccsanady@scl.ameslab.gov, hackers@freebsd.org, matt@3am-software.com
Subject: Re: Network concurrency problems!?
In-Reply-To: <6925.866703623@verdi.nethelp.no>
Message-ID: <Pine.BSF.3.95q.970619011114.12433E-100000@misery.sdf.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk


On Thu, 19 Jun 1997 sthaug@nethelp.no wrote:

> > > There has been a good deal of debate on whether offloading is really the
> > > best idea for network protocol implementations. A lot of people have tried
> > > it, and a lot of people have failed.  If you look at Van Jacobson't work
> > > you'll find him arguing in the opposite direction: A "stupid" (in reality:
> > > simple and efficient) controller, and a very efficient protocol stack
> > > implementation.
> > 
> >   Not really what I was refering to.  I was thinking about controllers
> > that minimize interupt calls, use DMA, avoid PIO, and align data
> > transfered to the host.
> 
> With the possible exception of the data alignment issue, which of these do
> you think are *not* used by the 100 Mbps cards supported by FreeBSD?

  Not "think", but "know".  The 3COM Fast Etherlink is only used in polled
mode.  Even then, the card does seem to be terribly efficient.  You should
read the driver, and card specs for more details, rather than bother the
list.

  Also, DG has mentioned that the Intel Etherexpress Pro100, is faster
than anything using the de driver, because the driver is simpler, due to
the hardware design of the Pro100.

  Anyways, enough on this topic.  See the list archives, and hw specs for
more details.

> Steinar Haug, Nethelp consulting, sthaug@nethelp.no
> 
> 

Tom