From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [140.186.70.92] (port=36315 helo=eggs.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1PmpBw-0000He-M5 for qemu-devel@nongnu.org; Tue, 08 Feb 2011 10:10:06 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1PmpBL-0006My-ME for qemu-devel@nongnu.org; Tue, 08 Feb 2011 10:08:45 -0500 Received: from hall.aurel32.net ([88.191.126.93]:58199) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1PmpBL-0006MG-EY for qemu-devel@nongnu.org; Tue, 08 Feb 2011 10:08:43 -0500 Message-ID: <4D515C6C.9000407@aurel32.net> Date: Tue, 08 Feb 2011 16:08:28 +0100 From: Aurelien Jarno MIME-Version: 1.0 Subject: Re: [Qemu-devel] Re: [PATCH 2/7] Enable I/O thread and VNC threads by default References: <4D3DFD20.8060004@linux.vnet.ibm.com> <20110125091741.GB30239@edde.se.axis.com> <20110125133453.GC5427@amt.cnet> <20110207101255.GA20413@amt.cnet> <20110207160350.GA26332@amt.cnet> <4D501C71.7090708@redhat.com> <4D50279B.5010102@siemens.com> <4D505DCB.9050406@codemonkey.ws> <20110207214551.GB16429@hall.aurel32.net> <4D50A5F0.802@codemonkey.ws> <20110208072657.GD16429@hall.aurel32.net> <4D50FA14.5010100@redhat.com> <4D5103E8.6050808@siemens.com> <4D510771.3040309@aurel32.net> <4D511221.9030505@siemens.com> <4D5113D3.9090802@aurel32.net> <4D511500.1040303@siemens.com> <4D5115C2.6060008@aurel32.net> <4D51842C.8000209@codemonkey.ws> <4D5125E2.8090902@aurel32.net> <4D5132C3.1040900@redhat.com> <4D5145A4.1070904@aurel32.net> In-Reply-To: <4D5145A4.1070904@aurel32.net> Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 8bit List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: Stefan Hajnoczi , Jan Kiszka , Marcelo Tosatti , "qemu-devel@nongnu.org" , Anthony Liguori , Paul Brook , Arun Bharadwaj , "Edgar E. Iglesias" Aurelien Jarno a écrit : > Paolo Bonzini a écrit : >> On 02/08/2011 12:15 PM, Aurelien Jarno wrote: >>> however >>> it should not be done ignoring all the*current* drawbacks of the >>> iothread mode. We know them (at least for some of them), so let's try to >>> solve them. >> Let's also enumerate them. >> > > From what I know: > - performance regression in TCG mode I setup an x86_64 guest on an x86_64 host (Intel Xeon E5345). Nothing was running except the standard daemons and the CPU governor was set to "performance" on all CPU. I then compared the network performance using netperf in default mode, through a tap interface and a virtio nic. I got the following results (quite reproducible, std below 0.5): - without IO thread: 107.36 MB/s - with IO thread: 89.93 MB/s I haven't redone the tests I have done two weeks ago on MIPS, ARM, PowerPC and SH4 (using different emulated network cards: smc91c111, rtl8139, e1000, virtio), but it was roughly the same slow down, except on ARM where it was more important. -- Aurelien Jarno GPG: 1024D/F1BCDB73 aurelien@aurel32.net http://www.aurel32.net