From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesper Dangaard Brouer Subject: Re: [PATCH 0/2] Get rid of ndo_xmit_flush Date: Wed, 27 Aug 2014 15:56:51 +0200 Message-ID: <20140827155651.04fe49d0@redhat.com> References: <20140825.163458.1117073971092495452.davem@davemloft.net> <1409142672.26515.24.camel@localhost> <1409145838.3173.41.camel@edumazet-glaptop2.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: Hannes Frederic Sowa , David Miller , netdev@vger.kernel.org, therbert@google.com, jhs@mojatatu.com, edumazet@google.com, jeffrey.t.kirsher@intel.com, rusty@rustcorp.com.au, dborkman@redhat.com, brouer@redhat.com To: Eric Dumazet Return-path: Received: from mx1.redhat.com ([209.132.183.28]:32760 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933404AbaH0N5H (ORCPT ); Wed, 27 Aug 2014 09:57:07 -0400 In-Reply-To: <1409145838.3173.41.camel@edumazet-glaptop2.roam.corp.google.com> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, 27 Aug 2014 06:23:58 -0700 Eric Dumazet wrote: > I am afraid Jesper tests are not complete to truly have a picture of the > extra costs, because he made sure no false sharing was possible (A > single cpu does everything and keeps skb in its cache) I fully agree. My tests a very artificial benchmarks only for the optimal case. I do worry a bit, if writing skb->xmit_more in a more cache cold scenarios could hurt us (as that cacheline seems to be read mostly), but this is not a problem right now because it will always be cache hot, as we always clear the entire SKB in alloc. -- Best regards, Jesper Dangaard Brouer MSc.CS, Sr. Network Kernel Developer at Red Hat Author of http://www.iptv-analyzer.org LinkedIn: http://www.linkedin.com/in/brouer