From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:41263) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VRzv4-0004qI-C3 for qemu-devel@nongnu.org; Fri, 04 Oct 2013 03:35:33 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1VRzuw-0006NA-OZ for qemu-devel@nongnu.org; Fri, 04 Oct 2013 03:35:26 -0400 Received: from moutng.kundenserver.de ([212.227.17.8]:54053) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1VRzuw-0006MY-Fh for qemu-devel@nongnu.org; Fri, 04 Oct 2013 03:35:18 -0400 Message-ID: <524E6FB1.30103@rdsoftware.de> Date: Fri, 04 Oct 2013 09:35:13 +0200 From: Erik Rull MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Subject: [Qemu-devel] CPU Benchmark qemu-kvm 1.2.0 Win XP vs. Win 7 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "qemu-devel@nongnu.org" Hi all, when testing my first Windows 7 guest, I noticed, that the IRQ load / kernel time within the guest is quite high when having network access. I use the virtio drivers - the same version as in the Windows XP guests. There the kernel load is nearly only 50% of the Win7 load. Why? The provided guest hardware is exactly the same... Win XP was used without ACPI, Win7 requires it - so it is enabled on both commandlines. But I can't imagine that this causes such a huge performance drop. Additionally I did some CPU benchmarking and the Win7 guest results in ~ 20% less in the hard/wetstone performance. Also the harddisk access is slower. Any ideas what may cause the pure hardware performance degradation between these two guests when having the same qemu-kvm hardware parameters (beside the image file of course)? Any hints what to do to improve the performance for Win7 would be appreciated. I would at least assume that the pure CPU performance would not drop significantly. Thanks. Best regards, Erik