From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1L8fhy-0008Ro-51 for qemu-devel@nongnu.org; Fri, 05 Dec 2008 13:47:22 -0500 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1L8fhx-0008RV-PY for qemu-devel@nongnu.org; Fri, 05 Dec 2008 13:47:21 -0500 Received: from [199.232.76.173] (port=60632 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1L8fhx-0008RN-Cl for qemu-devel@nongnu.org; Fri, 05 Dec 2008 13:47:21 -0500 Received: from mx2.redhat.com ([66.187.237.31]:59715) by monty-python.gnu.org with esmtp (Exim 4.60) (envelope-from ) id 1L8fhw-0001QC-Tf for qemu-devel@nongnu.org; Fri, 05 Dec 2008 13:47:21 -0500 Message-ID: <4939771A.1010602@redhat.com> Date: Fri, 05 Dec 2008 20:46:50 +0200 From: Avi Kivity MIME-Version: 1.0 Subject: Re: [Qemu-devel] [RFC] Coalesced MMIO support References: <49384758.9090900@us.ibm.com> In-Reply-To: <49384758.9090900@us.ibm.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit 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 Cc: kvm-devel , Laurent Vivier , Ian Jackson , Stefano Stabellini Anthony Liguori wrote: > The images I need to test this against are not on this computer, so I > can't commit until tomorrow when I have access to them again. So I > figured that was a good reason to post first as an RFC. > > MMIO exits are more expensive in KVM or Xen than in QEMU because they > involve, at least, privilege transitions. However, MMIO write > operations can be effectively batched if those writes do not have side > effects. > > Good examples of this include VGA pixel operations when in a planar > mode. As it turns out, we can get a nice boost in other areas too. > Laurent mentioned a 9.7% performance boost in iperf with the coalesced > MMIO changes for the e1000 when he originally posted this work for KVM. > The cirrus blitter could also use this. None of the registers except start/status have any effect except for queuing a bit of data. -- I have a truly marvellous patch that fixes the bug which this signature is too narrow to contain.