From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Ahern Subject: Re: performance of virtual functions compared to virtio Date: Mon, 25 Apr 2011 13:49:39 -0600 Message-ID: <4DB5D053.1070401@gmail.com> References: <4DAF8EF0.8010203@gmail.com> <1303353349.3110.181.camel@x201> <4DB5B1C4.4000602@gmail.com> <1303755193.3431.50.camel@x201> <4DB5C65C.20306@gmail.com> <1303759773.3431.64.camel@x201> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: KVM mailing list To: Alex Williamson Return-path: Received: from mail-pv0-f174.google.com ([74.125.83.174]:55341 "EHLO mail-pv0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758152Ab1DYTtm (ORCPT ); Mon, 25 Apr 2011 15:49:42 -0400 Received: by pvg12 with SMTP id 12so1365659pvg.19 for ; Mon, 25 Apr 2011 12:49:41 -0700 (PDT) In-Reply-To: <1303759773.3431.64.camel@x201> Sender: kvm-owner@vger.kernel.org List-ID: On 04/25/11 13:29, Alex Williamson wrote: > So we're effectively getting host-host latency/throughput for the VF, > it's just that in the 82576 implementation of SR-IOV, the VF takes a > latency hit that puts it pretty close to virtio. Unfortunate. I think For host-to-VM using VFs is worse than virtio which is counterintuitive. > you'll find that passing the PF to the guests should be pretty close to > that 185us latency. I would assume (hope) the higher end NICs reduce About that 185usec: do you know where the bottleneck is? It seems as if the packet is held in some queue waiting for an event/timeout before it is transmitted. David > this, but it seems to be a hardware limitation, so it's hard to predict. > Thanks, > > Alex >