From owner-freebsd-mobile  Wed Oct 15 06:51:15 1997
Return-Path: <owner-freebsd-mobile>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id GAA15486
          for mobile-outgoing; Wed, 15 Oct 1997 06:51:15 -0700 (PDT)
          (envelope-from owner-freebsd-mobile)
Received: from word.smith.net.au (ppp20.portal.net.au [202.12.71.120])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id GAA15435
          for <freebsd-mobile@FreeBSD.ORG>; Wed, 15 Oct 1997 06:50:17 -0700 (PDT)
          (envelope-from mike@word.smith.net.au)
Received: from word.smith.net.au (localhost [127.0.0.1])
	by word.smith.net.au (8.8.7/8.8.5) with ESMTP id XAA00752;
	Wed, 15 Oct 1997 23:15:57 +0930 (CST)
Message-Id: <199710151345.XAA00752@word.smith.net.au>
X-Mailer: exmh version 2.0zeta 7/24/97
To: Ben <ben@edelweb.fr>
cc: Mike Smith <mike@smith.net.au>, Adept <adept@cep.yale.edu>,
        Brian Somers <brian@awfulhak.org>, William Bulley <web@merit.edu>,
        freebsd-mobile@FreeBSD.ORG
Subject: Re: 3C589 performance (was Re: obtaining 3COM 3C589C PC-CARDs ) 
In-reply-to: Your message of "Wed, 15 Oct 1997 15:23:05 +0200."
             <Pine.BSF.3.91.971015152201.12107D-100000@ben.edelweb.fr> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Wed, 15 Oct 1997 23:15:56 +0930
From: Mike Smith <mike@smith.net.au>
Sender: owner-freebsd-mobile@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> > > I 've done some benchmarking and I'm getting on my Thinkpad 701C something
> > > like 700K/sec with my 3C589B.
> > 
> > On a silent network, their throughput is fine.  Unfortunately, it 
> > appears that the '589 suffers from similar problems to other 3Com cards 
> > on networks with other traffic; I see transfers out of this system as 
> > slow as 50K/sec on a lightly to moderately loaded network.  Others have 
> > reported similar experiences with other cards (3C509, 3C59x, 3C90x).
> 
> I have had this problem before--what you have to make sure ofis:
> 
> 1)	The cards aren't sharing interrupts

You can't share interrupts with a PCCARD.

> 2)	You aren't using more than one link state for the card at a time
> 	(i.e. -link0 link1 -link2)

ep0: flags=a843<UP,BROADCAST,RUNNING,SIMPLEX,LINK1,MULTICAST> mtu 1500
        inet 202.0.75.17 netmask 0xfffffff0 broadcast 202.0.75.31
        ether 00:60:97:88:e7:93 

On the same network, talking to the same client, an 
NE2000-clone-equipped laptop achieves around 700k/sec.   I'm looking 
forward to the changes recently committed to the if_ep driver in the 
hope that this will help, although I am about to lose my 589 when I 
have to give the cable back.

mike