From mboxrd@z Thu Jan 1 00:00:00 1970 From: Kumar Sanghvi Subject: Re: [PATCH net-next] cxgb4: Add low latency socket busy_poll support Date: Wed, 4 Feb 2015 12:24:37 +0600 Message-ID: <20150204062436.GE30631@kumar-pc.asicdesigners.com> References: <1422940816-27402-1-git-send-email-hariprasad@chelsio.com> <54D0FB41.1030802@hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Cc: Hariprasad Shenai , , , , , , To: Rick Jones Return-path: Received: from stargate.chelsio.com ([67.207.112.58]:24300 "EHLO stargate.chelsio.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751321AbbBDGyp (ORCPT ); Wed, 4 Feb 2015 01:54:45 -0500 Content-Disposition: inline In-Reply-To: <54D0FB41.1030802@hp.com> Sender: netdev-owner@vger.kernel.org List-ID: Hi Rick, On Tuesday, February 02/03/15, 2015 at 08:45:53 -0800, Rick Jones wrote: > On 02/02/2015 09:20 PM, Hariprasad Shenai wrote: > >cxgb_busy_poll, corresponding to ndo_busy_poll, gets called by the socket > >waiting for data. > > > >With busy_poll enabled, improvement is seen in latency numbers as observed by > >collecting netperf TCP_RR numbers. > > Which, not surprisingly, leads to the question - how much > improvement? :) Extra credit for including the change in > netperf-reported service demands. > > happy benchmarking, > > rick jones > netperf was not changed in this case. Rather, busy-polling was enabled globally by setting sysctls busy_poll and busy_read. We will post the improvement numbers when submitting V2. Thanks, Kumar.