From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: setsockopt() Date: Tue, 8 Jul 2008 13:48:45 -0700 Message-ID: <20080708134845.2372a483@speedy> References: <48725DFE.6000504@citi.umich.edu> <20080707142408.43aa2a2e@extreme> <48728B09.1050801@citi.umich.edu> <20080707.144912.76654646.davem@davemloft.net> <20080708045443.GA7726@2ka.mipt.ru> <20080708020235.388a7bd5.billfink@mindspring.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: Bill Fink , Evgeniy Polyakov , David Miller , aglo@citi.umich.edu, shemminger@vyatta.com, netdev@vger.kernel.org, rees@umich.edu, bfields@fieldses.org To: Roland Dreier Return-path: Received: from mail.vyatta.com ([216.93.170.194]:49386 "EHLO mail.vyatta.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751066AbYGHUsr (ORCPT ); Tue, 8 Jul 2008 16:48:47 -0400 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Mon, 07 Jul 2008 23:29:31 -0700 Roland Dreier wrote: > Interesting... I'd not tried nuttcp before, and on my testbed, which is > a very high-bandwidth, low-RTT network (IP-over-InfiniBand with DDR IB, > so the network is capable of 16 Gbps, and the RTT is ~25 microseconds), > the difference between autotuning and not for nuttcp is huge (testing > with 2.6.26-rc8 plus some pending 2.6.27 patches that add checksum > offload, LSO and LRO to the IP-over-IB driver): > > nuttcp -T30 -i1 ends up with: > > 14465.0625 MB / 30.01 sec = 4043.6073 Mbps 82 %TX 2 %RX > > while setting the window even to 128 KB with > nuttcp -w128k -T30 -i1 ends up with: > > 36416.8125 MB / 30.00 sec = 10182.8137 Mbps 90 %TX 96 %RX > > so it's a factor of 2.5 with nuttcp. I've never seen other apps behave > like that -- for example NPtcp (netpipe) only gets slower when > explicitly setting the window size. > > Strange... I suspect that the link is so fast that the window growth isn't happening fast enough. With only a 30 second test, you probably barely made it out of TCP slow start.