From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1DviWX-0006VO-Im for qemu-devel@nongnu.org; Thu, 21 Jul 2005 17:24:11 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1DviWO-0006QN-9V for qemu-devel@nongnu.org; Thu, 21 Jul 2005 17:24:01 -0400 Received: from [199.232.76.173] (helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1DviWN-0006LQ-9L for qemu-devel@nongnu.org; Thu, 21 Jul 2005 17:23:59 -0400 Received: from [84.96.92.60] (helo=Smtp.neuf.fr) by monty-python.gnu.org with esmtp (Exim 4.34) id 1DviWV-0007Qv-EZ for qemu-devel@nongnu.org; Thu, 21 Jul 2005 17:24:07 -0400 Received: from [84.99.204.52] by sp604001mt.gpm.neuf.ld (Sun Java System Messaging Server 6.2-2.05 (built Apr 28 2005)) with ESMTP id <0IJZ00IL0XNG2I20@sp604001mt.gpm.neuf.ld> for qemu-devel@nongnu.org; Thu, 21 Jul 2005 23:14:04 +0200 (CEST) Date: Thu, 21 Jul 2005 23:14:32 +0200 From: Fabrice Bellard Subject: Re: [Qemu-devel] Can you commit the pcnet patch ? In-reply-to: <200507211419.51421.paul@codesourcery.com> Message-id: <42E01038.8030202@bellard.org> MIME-version: 1.0 Content-type: text/plain; charset=us-ascii; format=flowed Content-transfer-encoding: 7BIT References: <42DE282A.3080102@linbox.com> <20050721072943.GA24642@tuxedo.skovlyporten.dk> <200507211419.51421.paul@codesourcery.com> Reply-To: qemu-devel@nongnu.org List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Paul Brook wrote: > On Thursday 21 July 2005 08:29, Lars Munch wrote: > >>On Wed, Jul 20, 2005 at 12:32:10PM +0200, Ludovic Drolez wrote: >> >>>Hi ! >>> >>>I've tried the latest AMD pcnet patch (made by Antony T Curtis) with qemu >>>0.7 because I was tired of the sloooww PXE booting of my QEMU. For >>>exemple, to load a linux kernel + initrd over TFTP, it takes 4m30s with a >>>ne2000, and 1m35s with the PCNET card. >> >>Does anyone know why pcnet is so much faster than ne2000? is it easier >>to emulate? or is there room for improvements on the ne2000? > > > Probably because it's a DMA capable controller. IIRC the ne2k is pio based, so > has to shuffle the data through an IO port one byte at a time. I don't think that the switch to DMA can explain a factor of 3. Another possibility is that there is a bug in the ne2000 code. It could be interesting to check the send and receive latencies by adding traces in the ne2000 code... Fabrice.