From owner-freebsd-current Sun Sep 14 13:51:42 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id NAA29826 for current-outgoing; Sun, 14 Sep 1997 13:51:42 -0700 (PDT) Received: from duke.cs.duke.edu (duke.cs.duke.edu [152.3.140.1]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id NAA29813; Sun, 14 Sep 1997 13:51:30 -0700 (PDT) Received: from hurricane.cs.duke.edu (hurricane.cs.duke.edu [152.3.145.1]) by duke.cs.duke.edu (8.8.5/8.8.5) with ESMTP id QAA01316; Sun, 14 Sep 1997 16:51:26 -0400 (EDT) Received: (from gallatin@localhost) by hurricane.cs.duke.edu (8.8.4/8.7.3) id QAA27413; Sun, 14 Sep 1997 16:51:26 -0400 (EDT) Date: Sun, 14 Sep 1997 16:51:26 -0400 (EDT) Message-Id: <199709142051.QAA27413@hurricane.cs.duke.edu> From: Andrew Gallatin MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: "Gary Palmer" Cc: Karl Denninger , current@FreeBSD.ORG Subject: Re: Problems? In-Reply-To: <7386.874113565@orion.webspan.net> References: <19970912094231.48481@Jupiter.Mcs.Net> <7386.874113565@orion.webspan.net> X-Mailer: VM 6.32 under 19.15 XEmacs Lucid Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Gary Palmer writes: > Karl Denninger wrote in message ID > <19970912094231.48481@Jupiter.Mcs.Net>: > > Hi foolks, > > > > Anyone got an idea what this means? > > > > de1: abnormal interrupt: transmit underflow (raising TX threshold to 96|256) > > de1: abnormal interrupt: transmit underflow (raising TX threshold to 8|512) > > de1: abnormal interrupt: transmit underflow (raising TX threshold to 1024) > > de1: abnormal interrupt: transmit underflow (switching to store-and-forward > > mode > > According to Matt, this is a problem with the PCI bus on the m/b not > being fast enough to handle the data. He says the Natoma is > notoriously slow (which is why I saw the messages too) Actually, the the Natoma is pretty nice. Under heavy load, its true that it might not get DMA's done as quickly as a Triton. But that's not because the Natoma is slow, rather its because the the Natoma is fair. Under heavy load, a Triton will starve the CPU from memory, whereas a Natama will divide the memory bandwidth up pretty much equally between the PCI bus and the CPU. I imagine this would allow the host to post more transmits & make the card fall further behind.. Drew ------------------------------------------------------------------------------ Andrew Gallatin, Sr Systems Programmer http://www.cs.duke.edu/~gallatin Duke University Email: gallatin@cs.duke.edu Department of Computer Science Phone: (919) 660-6590