Return-path: Received: from mail-we0-f169.google.com ([74.125.82.169]:51248 "EHLO mail-we0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751339AbaIRSrt convert rfc822-to-8bit (ORCPT ); Thu, 18 Sep 2014 14:47:49 -0400 Received: by mail-we0-f169.google.com with SMTP id w61so1401599wes.14 for ; Thu, 18 Sep 2014 11:47:48 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <773DB8A82AB6A046AE0195C68612A319018C1FBC@sbs2003.acksys.local> References: <773DB8A82AB6A046AE0195C68612A319018C1FBC@sbs2003.acksys.local> Date: Thu, 18 Sep 2014 20:47:48 +0200 Message-ID: (sfid-20140918_204753_100677_4FEA1DD8) Subject: Re: [PATCHv2 00/10] add support for ACK timeout estimation in ath9k driver From: Lorenzo Bianconi To: Cedric Debarge Cc: linux-wireless Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: > On Tue, Sept 16, 2014 at 02:13:25AM +0200, Lorenzo Bianconi wrote: > > Hi, > Hi, >> This patchset adds support for estimation of the ACK timeout (dynack) in ath9k driver. Ath9k dynack computes the > ACK timeout based on ACK frame RX timestamp, TX frame timestamp and frame duration. >> >> Ath9k dynack has been tested in indoor environment using AR9223/AR9280 chipset (running 3.17.0-rc5 kernel), on >> 9Km PtoP link using AR9280 chipset (running OpenWRT trunk, compat-wireless-2014-05-22) and on an AP using AR9280 > chipset (running OpenWRT trunk, compat-wireless-2014-05-22) serving 35 STA with links up to 8Km > > During your indoor tests, did you managed to get precise enough timestamps to compute a significant ACK to (despite of short distances and multipaths)? > Dynack tx/rx timestamps are snapshots of PCU's timestamps (expressed in TSF value). > Previous tests have shown impressive bandwith gains (10% on a 9Km P2P link). Did you also get improvements in indoor environment? > In indoor environment I got the same performance obtained with a static ACK timeout since LER was roughly 0%. Anyway I guess dynack should be enabled just on outdoor "long" links (like ones tested by Philippe) since that environment is really sensitive to ACK timeout. > Thanks, > > Cédric DEBARGE Regards, Lorenzo -- UNIX is Sexy: who | grep -i blonde | talk; cd ~; wine; talk; touch; unzip; touch; strip; gasp; finger; gasp; mount; fsck; more; yes; gasp; umount; make clean; sleep