Re: [Madwifi-users] performance and reliability problems - solutions on the horizon? (madwifi-bsd, hostapd wpa)

[Andargor T. W. <and...@ya...>]

OpenWrt includes bwm (bandwidth monitor). I'm not sure
if it's as accurate as iperf, but it reports both tx
and rx at ~2000 KBps on the WRT54GS. Firefox, for the
same transfer, reported 1875 KBps. I'm not sure if
both of these strip out protocol overhead (HTTP
transfer).

OpenWrt is basically Linux, so I'll see if there's an
iperf package for it.

Andargor

--- Ethan Tuttle <et...@et...> wrote:

> I've been using iperf.  I run it as a server on the
> madwifi ap, and I run 
> 'iperf -c ap_hostname -r' on the client.  I like
> iperf because it gives the 
> effective throughput (after tcp overhead) in each
> direction.  I also keep an 
> eye on the AP's tx rate as reported in 
> /proc/sys/net/wlan/ath0/associated_sta.
> 
> My experience is much better since I switched to
> sample rate control this 
> morning.  It seems stable now!  I suppose I
> underestimated the importance of 
> rate control, and how much better sample could be. 
> From my experience: 
> sample = great, onoe = sorta works, aamr = short
> range only.
> 
> It still seems like the AP could transmit faster.
> Unfortunately, I don't 
> have another 11g access point to compare against.
> 
> Ethan
> 
> ----- Original Message ----- 
> From: "Andargor The Wise" <and...@ya...>
> To: "Ethan Tuttle" <et...@et...>;
> "madwifi-users" 
> <mad...@li...>
> Sent: Saturday, July 02, 2005 2:19 PM
> Subject: Re: [Madwifi-users] performance and
> reliability problems - 
> solutions on the horizon? (madwifi-bsd, hostapd wpa)
> 
> 
> > What tool are you using to measure bandwidth to
> get
> > the tx/rx distinction? Are you simply downloading
> and
> > then uploading, and then looking at the transfer
> rate
> > in each direction?
> >
> > I'm asking because I can do some comparisons with
> > masdwifi and other APs. I've got a D-Link DI-624,
> and
> > I've recently acquired a Linksys WRT54GS which
> I've
> > loaded with OpenWrt (very impressive, BTW).
> >
> > I've done performance comparisons before with the
> > DI-624 here:
> >
> > http://www.andargor.com/img/di-624.gif
> > http://www.andargor.com/img/madwifi-txfer.gif
> > http://www.andargor.com/img/error-comparison2.gif
> >
> > This was with onoe, sample gives me something
> similar
> > to the DI-624.
> >
> > Andargor
> >
> >
> > --- Ethan Tuttle <et...@et...> wrote:
> >
> >> At Andargor's suggestion, I switched from onoe to
> >> sample rate control.  This
> >> is after a bad experience with amrr rate control
> -
> >> amrr appeared to keep the
> >> AP from retraining effectively to support my
> usual
> >> work location one floor
> >> away.
> >>
> >> Initial impressions of sample are good.  It is
> >> retraining to higher rates
> >> without killing connections and dropping packets.
> >>
> >> I'm still seeing a large gap between the tx rate
> of
> >> the AP and the tx rate
> >> of my XP client.  At the moment, here's what
> iperf
> >> is showing:
> >>
> >> rx: 4.18 Mbits/sec (XP client with Intel 2200BG
> >> sending)
> >> tx: 1.22 Mbits/sec (AP with madwifi sending)
> >>
> >> Both endpoints rapidly change their tx rates,
> >> especially in response to
> >> packet storms like iperf.  Before running the
> above,
> >> the client was at
> >> 11Mbps and after it was at 1Mbps.  The AP was
> >> sending at 48Mbps before and
> >> 5.5Mbps after.  And the client still got much
> better
> >> tx throughput!  I do
> >> see the throughput numbers fluctuate quite a bit,
> >> but the AP always
> >> transmits at a lower rate than the client.
> >>
> >> These numbers reinforce my observation that
> madwifi
> >> has transmit problems.
> >> Anyone got an idea why?  I want to believe that
> it
> >> will be improved.
> >> However I haven't seen the madwifi developers
> >> acknowledge the problem!
> >>
> >> Ethan
> >>
> >> PS I should mention that my AP has a high-gain
> >> omnidirectional antenna,
> >> while the client is a laptop with internal
> antenna.
> >> I suppose that would
> >> give the client an easier time of transmitting,
> >> because the AP is more
> >> sensitive when it comes to receiving.  But I am
> >> still suspicious of such a
> >> large discrepancy in the throughput speeds.
> >> Suggestions on how to account
> >> for the different antennas in my tests would be
> >> appreciated.
> >>
> >> ----- Original Message ----- 
> >> From: "Andargor The Wise" <and...@ya...>
> >> To: "Ethan Tuttle" <et...@et...>;
> >> "madwifi-users"
> >> <mad...@li...>
> >> Sent: Friday, July 01, 2005 12:51 PM
> >> Subject: Re: [Madwifi-users] performance and
> >> reliability problems -
> >> solutions on the horizon? (madwifi-bsd, hostapd
> wpa)
> >>
> >>
> >> FWIW, I've found better performance and stability
> >> using sample instead of onoe or amrr.
> >>
> >> I haven't done performance testing at my
> network's
> >> edge, however, but 40 feet away and through a
> wall I
> >> get ~1,600KBps (12Mbps) throughput. I would
> normally
> >> expect around 20-22 Mbps actual throughput from
> >> straight 11g.
> >>
> >> My configuration:
> >>
> >> http://www.andargor.com/files/madwifi-info.txt
> >>
> >> Andargor
> >>
> >> --- Ethan Tuttle <et...@et...> wrote:
> >>
> >> > I'm having performance and reliability problems
> >> with
> >> > my madwifi based
> >> > 802.11g access point.  I've been researching
> this
> >> > but I'm stuck.
> >> >
> >> > I first noticed that the rx rate on my XP
> client
> >> was
> >> > much lower than the tx
> >> > rate.  In my usual work location (one floor up
> >> from
> >> > the AP), I was receiving
> >> > at 80KBps while I cound send at 220KBps.  I
> >> figured
> >> > out that this was
> >> > because madwifi had decreased it's tx rate,
> while
> >> my
> >> > XP client was happy to
> >> > transmit faster. (!)  With my old 11b bridge, I
> >> had
> >> > stable 150KBps
> >> > connections at this same location.
> >> >
> >> > Sitting next to the AP, I measured TCP
> throughput
> >> at
> >> > 2.25 MBps round-trip.
> >> > Which was not bad, as long as it lasted.  Over
> >> time
> >> > the AP's tx rate would
> >> > drop to 9Mbit or lower, and it wouldn't ever
> >> > recover.
> >> >
> >> > I thought it would be clever to switch from
> onoe
> >> to
> 
=== message truncated ===



		
__________________________________ 
Discover Yahoo! 
Find restaurants, movies, travel and more fun for the weekend. Check it out! 
http://discover.yahoo.com/weekend.html