From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mtiwmhc13.worldnet.att.net ([204.127.131.117]:58005 "EHLO mtiwmhc13.worldnet.att.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751866AbYGZPyx (ORCPT ); Sat, 26 Jul 2008 11:54:53 -0400 Message-ID: <488B48C1.6090506@lwfinger.net> (sfid-20080726_175456_991253_69591B25) Date: Sat, 26 Jul 2008 10:54:41 -0500 From: Larry Finger MIME-Version: 1.0 To: Felix Fietkau CC: =?ISO-8859-1?Q?C=E9dric_Auger?= , linux-wireless@vger.kernel.org Subject: Re: iwl3945 slow for downloading References: <20080726161140.59041730@grisbi.org> <488B3969.4050405@openwrt.org> In-Reply-To: <488B3969.4050405@openwrt.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Sender: linux-wireless-owner@vger.kernel.org List-ID: Felix Fietkau wrote: > I don't know if it helps with debugging this issue, but I've seen the same > on a laptop from a friend of mine (same card). > > After some searching I found that setting the AP to g-only mode instead of > b+g mixed made it work at decent speeds. Another symptom of this slowdown is > that there is typically an unnaturally high number of retransmissions on the > air, which eats up a lot of airtime. > > I've spent some time looking through the code to find the reason for this > but couldn't find anything yet (maybe somebody from Intel could help). Just for comparison, my BCM4312 running b43 from the latest wireless-testing gets the following tcpperf throughput: AP Mode TX RX Mixed b/g 19.8 MB/s 23.0 MB/s g-only 20.5 MB/s 23.1 MB/s For these tests, the connecting end was wired to the router. The difference in TX speeds was about what I expected - the small difference in RX speeds was not. In any case, if you see a large difference with iwl3945, that would be useful information. You can obtain tcpperf from http://wand.cs.waikato.ac.nz/~stj2/nsc/software.html Larry