Return-path: Received: from vms173017pub.verizon.net ([206.46.173.17]:58485 "EHLO vms173017pub.verizon.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751738AbZBCEYN (ORCPT ); Mon, 2 Feb 2009 23:24:13 -0500 Received: from depotrd.localnet ([96.237.24.216]) by vms173017.mailsrvcs.net (Sun Java(tm) System Messaging Server 6.3-2.01 (built Jun 13 2007; 32bit)) with ESMTPA id <0KEG00MBHXVSCDY8@vms173017.mailsrvcs.net> for linux-wireless@vger.kernel.org; Mon, 02 Feb 2009 20:35:53 -0600 (CST) From: Glenn Burkhardt To: Nick Kossifidis Subject: Re: poor performance with ath5k; reverting to ath_pci Date: Mon, 02 Feb 2009 21:36:00 -0500 References: <200901140003.31693.gbburkhardt@verizon.net> <200901212158.13854.gbburkhardt@verizon.net> <40f31dec0901222331l2d55b732ua50089aef8a42ccf@mail.gmail.com> In-reply-to: <40f31dec0901222331l2d55b732ua50089aef8a42ccf@mail.gmail.com> Cc: linux-wireless@vger.kernel.org MIME-version: 1.0 Content-type: text/plain; charset=utf-8 Message-id: <200902022136.00606.gbburkhardt@verizon.net> (sfid-20090203_052417_963344_FB67FFCB) Sender: linux-wireless-owner@vger.kernel.org List-ID: > O.K. so both algorithms have overestimated the rate, madwifi with > minstrel sets it to 36M and both pid an minstrel on ath5k/mac80211 set > it to 54M. This results packet corruption and packet loss. Please try > locking the rate to 18M - 36M and check the performance again (also an > iperf udp test would be very usefull) to see if this is only a problem > with the algorithm or a more generic problem. It doesn't seem to be the rate selection. I switched to channel 2 again, and the version of ath5k that came with Suse 11.1 didn't connect to my router at all. The ath_pci driver I built did, with roughly millisecond ping round trips. The ath_pci driver changed the bit rate to 18 Mb/s; ath5k reported using 5.5 Mb/s, but ping was never successful. I also built a kernel from the wireless testing repository ( 2.6.29-rc3-wl-9- pae), and that driver was able to connect initially, although with a slow response rate: PING 192.168.2.1 (192.168.2.1) 56(84) bytes of data. 64 bytes from 192.168.2.1: icmp_seq=1 ttl=64 time=3086 ms 64 bytes from 192.168.2.1: icmp_seq=2 ttl=64 time=2925 ms 64 bytes from 192.168.2.1: icmp_seq=3 ttl=64 time=3293 ms 64 bytes from 192.168.2.1: icmp_seq=4 ttl=64 time=3986 ms 64 bytes from 192.168.2.1: icmp_seq=6 ttl=64 time=4018 ms 64 bytes from 192.168.2.1: icmp_seq=7 ttl=64 time=3030 ms ping: sendmsg: Network is unreachable ping: sendmsg: Network is unreachable It reported the bit rate to be 54 Mb/s; I tried turning it down to as low as 6 Mb/s, but then it didn't work at all. This last test was using 'minstrel' rate control; it doesn't seem that 'pid' is a selectable option in the kernel configuration for 'wireless testing'. I'll try to get some more data with 'iperf'. I have a couple of other machines in the house that I can use.