Solved but puzzled: high cpu usage with IBM 802.11abg card

My Thinkpad T40 has IBM's 802.11a/b/g wireless card and this connects
to the downstairs router, a DLink G604 using WPA and TKIP.

On several occasions I have observed laptop CPU usage suddenly climb to
around 100% and then stay there indefinitely, and, after installing the
latest drivers and BIOS, this still wasn't fixed. If you disabled the
wireless card then CPU usage dropped right back down to zero again.
Re-enable it and up went the CPU.

After a good deal of headscratching I found that if I wandered
downstairs with the laptop and put it close to the router, suddenly CPU
usage dropped back down to zero. But the signal strength upstairs was
shown as excellent and the transmit rate was 54Mbps. So what was going
on?

I went back upstairs and the CPU usage then climbed back up again. So I
did a network scan and found that, hmm, there are three other networks
in the area, each with signal strength around 80%. Now, when I first
set everything up there was only my network when I scanned.

So then I thought, I wonder what happens if I select 802.11b transmit
mode on the card?. Well, this fixed the problem, albeit with a transfer
rate now set to 11Mbps. This isn't terribly much of a problem as
obviously my broadband link to the outside world is slower than that,
but this is a very interesting discovery. If you get a problem like
this and your card supports an alternative mode compatible with the
router, it might be worth trying that mode.

My suspicion is that there is a problem with the driver. When there are
multiple networks it starts doing a flurry of channel hopping and
signal strength checking - the diagnostic logs seem to support this
theory - and this chews up CPU like crazy. This might affect other
Atheros-based cards, as well.

<(E-Mail Removed)> wrote in message
news:(E-Mail Removed) ups.com...
> My Thinkpad T40 has IBM's 802.11a/b/g wireless card and this connects
> to the downstairs router, a DLink G604 using WPA and TKIP.
>
> On several occasions I have observed laptop CPU usage suddenly climb to
> around 100% and then stay there indefinitely, and, after installing the
> latest drivers and BIOS, this still wasn't fixed. If you disabled the
> wireless card then CPU usage dropped right back down to zero again.
> Re-enable it and up went the CPU.
>
> After a good deal of headscratching I found that if I wandered
> downstairs with the laptop and put it close to the router, suddenly CPU
> usage dropped back down to zero. But the signal strength upstairs was
> shown as excellent and the transmit rate was 54Mbps. So what was going
> on?
>
> I went back upstairs and the CPU usage then climbed back up again. So I
> did a network scan and found that, hmm, there are three other networks
> in the area, each with signal strength around 80%. Now, when I first
> set everything up there was only my network when I scanned.
>
> So then I thought, I wonder what happens if I select 802.11b transmit
> mode on the card?. Well, this fixed the problem, albeit with a transfer
> rate now set to 11Mbps. This isn't terribly much of a problem as
> obviously my broadband link to the outside world is slower than that,
> but this is a very interesting discovery. If you get a problem like
> this and your card supports an alternative mode compatible with the
> router, it might be worth trying that mode.
>
> My suspicion is that there is a problem with the driver. When there are
> multiple networks it starts doing a flurry of channel hopping and
> signal strength checking - the diagnostic logs seem to support this
> theory - and this chews up CPU like crazy. This might affect other
> Atheros-based cards, as well.
>
I have observed an almost identical situation using a USB adaptor with a
Prime RT2500 chipset. (PESI WU233g). Oddly, the AP is also a Dlink 604.
Unfortunately my location can "see" my AP and the annoying AP which is
usually running streaming video most of the time. He also runs a power
amplifier (big mouth, small ears) and his system is unaware of any other
user. Although we have locked down our transmit speeds to 2 Mbs, the receive
side appears to recognise all speeds. This probably is because the receive
side is clocked from the incoming received data stream. It would appear as
if drivers were written and tested in "clean" single user environments!

Peter

"Pierre" <(E-Mail Removed)> hath wroth:

>I have observed an almost identical situation using a USB adaptor with a
>Prime RT2500 chipset. (PESI WU233g). Oddly, the AP is also a Dlink 604.

The Dlink DI-604 is not a wireless router. I think you mean DI-514,
524, 624 or 824.

>Unfortunately my location can "see" my AP and the annoying AP which is
>usually running streaming video most of the time. He also runs a power
>amplifier (big mouth, small ears) and his system is unaware of any other
>user.

Lovely. Big mouth and small ears is called an "alligator".

>Although we have locked down our transmit speeds to 2 Mbs, the receive
>side appears to recognise all speeds. This probably is because the receive
>side is clocked from the incoming received data stream. It would appear as
>if drivers were written and tested in "clean" single user environments!

Nope. It's because *ALL* 802.11b management (and flow control) frames
are sent at the slowest 1Mbit/sec speed. That's to insure that
everyone can hear and decode everyone else regardless of their speed
settings.

This has nothing to do with the high CPU usage problem.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

"Jeff Liebermann" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> "Pierre" <(E-Mail Removed)> hath wroth:
>
> >I have observed an almost identical situation using a USB adaptor with a
> >Prime RT2500 chipset. (PESI WU233g). Oddly, the AP is also a Dlink 604.
>
> The Dlink DI-604 is not a wireless router. I think you mean DI-514,
> 524, 624 or 824.
>
The Dlink DSL-G604T is a ADSL Wireless Router/Modem
http://www.dlink.com.au/Default.aspx?ArticleID=740

> >Unfortunately my location can "see" my AP and the annoying AP which is
> >usually running streaming video most of the time. He also runs a power
> >amplifier (big mouth, small ears) and his system is unaware of any other
> >user.
>
> Lovely. Big mouth and small ears is called an "alligator".
>
> >Although we have locked down our transmit speeds to 2 Mbs, the receive
> >side appears to recognise all speeds. This probably is because the
receive
> >side is clocked from the incoming received data stream. It would appear
as
> >if drivers were written and tested in "clean" single user environments!
>
> Nope. It's because *ALL* 802.11b management (and flow control) frames
> are sent at the slowest 1Mbit/sec speed. That's to insure that
> everyone can hear and decode everyone else regardless of their speed
> settings.
>
> This has nothing to do with the high CPU usage problem.
>
> --
> Jeff Liebermann (E-Mail Removed)
> 150 Felker St #D http://www.LearnByDestroying.com
> Santa Cruz CA 95060 http://802.11junk.com
> Skype: JeffLiebermann AE6KS 831-336-2558

The USB port to the adaptor is working overtime during this high CPU
utilisation period. I should get some more tools in place and investigate
further but my time is now being spent building a small doubler from the
23cm gear that will quieten the local small eared alligator. The other
licence that Jeff and I have is quite handy!
Peter de C21TA et al

On Sat, 21 Jan 2006 11:09:54 +1100, "Pierre" <(E-Mail Removed)>
wrote:

>The Dlink DSL-G604T is a ADSL Wireless Router/Modem
>http://www.dlink.com.au/Default.aspx?ArticleID=740

The DI-604 is not wireless:
http://www.dlink.com/products/?pid=62
Your DSL-G604T is not listed on the USA site which I guess means that
it's not sold in the USA.

>The USB port to the adaptor is working overtime during this high CPU
>utilisation period.

Try running the Windoze Task Manager "Processes" tab.
Add to the columns with: View -> Columns
and add:
i/o reads
i/o writes
i/o other
They should show which program has gone nuts. Also, watch the CPU and
memory utilization.

>I should get some more tools in place and investigate
>further but my time is now being spent building a small doubler from the
>23cm gear that will quieten the local small eared alligator. The other
>licence that Jeff and I have is quite handy!
>Peter de C21TA et al

Don't bother. The carrier to interference ratio (jamming margin) on
spread sprectrum is about 10dB. See table at:

http://www.commsdesign.com/design_co...cleID=16505876
No way are you going to belch enough CW (or FM) to trash his receiver.
That's the whole idea behind spread spectrum. It's almost totally
immune to CW jamming. The best you can do is overload his receiver
(blocking). If you modulate your carrier with PN (pseudo-noise), you
might simulate an DSSS signal, which would certainly cause your victim
a problem. Now, if you really want power, how about just connecting a
coax cable to a microwave oven output coupleing loop, and really cook
them?



--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 (E-Mail Removed)
# http://802.11junk.com (E-Mail Removed)
# http://www.LearnByDestroying.com AE6KS

Jeff Liebermann <(E-Mail Removed)> wrote:
>Now, if you really want power, how about just connecting a
>coax cable to a microwave oven output coupleing loop, and really cook
>them?

Anyone else remember that guy who built a 2.4GHz amp from a microwave
oven klystron? He had a giant circulator so he could feed in a
modulating signal, and the klystron would lock onto it. It wasn't
clear from the article if he had yet fired it up, but the power supply
was truly scary!

William P.N. Smith <(E-Mail Removed)> hath wroth:

>Jeff Liebermann <(E-Mail Removed)> wrote:
>>Now, if you really want power, how about just connecting a
>>coax cable to a microwave oven output coupleing loop, and really cook
>>them?

>Anyone else remember that guy who built a 2.4GHz amp from a microwave
>oven klystron?

Microwave ovens use a magnetron, not a klystron. I went looking for
the article on converting a microwave oven into a transmitter. I
couldn't find it. Lots of articles on dangerous experiments (plasma
generators, ball lightning, sparks, etc) but nothing on use as a
communications device.

>He had a giant circulator so he could feed in a
>modulating signal, and the klystron would lock onto it. It wasn't
>clear from the article if he had yet fired it up, but the power supply
>was truly scary!

Well, let's do a bit of math. Let's pretend his antenna has a VSWR of
about 1.5:1 which will reflect about 5% of the power at the antenna.
The circulator dummy load has to absorb this reflected power. 800
watts coming from the magnetron yields 40 watts in the dummy load. 40
watts of dissipation will require a rather large dummy load. In
addition, the circulator usually burns about 1dB of loss, so it will
need to dissipate about 80 watts itself. I could probably build a
hybrid ring circulator that would handle that, but not a ferrite
circulator. How one would use a circulator to modulate a magnetron is
beyond my imagination. However, there are articles on phase locking
magentrons for use in a particle accellerator, so I guess it can be
done by phase locking it to an 802.11 source.

http://www.lancs.ac.uk/cockcroft-ins...ul04/tahir.pdf

Let me know if you find the construction article. I do remember
seeing (and ignoring) it about a year ago.


--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> wrote:
>William P.N. Smith <(E-Mail Removed)> hath wroth:

>>Anyone else remember that guy who built a 2.4GHz amp from a microwave
>>oven klystron?

>Microwave ovens use a magnetron, not a klystron.

[Oops, slept since then, and had kids, and all that.]

>Let me know if you find the construction article. I do remember
>seeing (and ignoring) it about a year ago.

This had to be back in my serious ham days (before kids), and since my
daughter is 12, it was at least a decade ago, well before everything
was on the WWWeb.

I just checked my stash of ham publications, and it's not there. IIRC,
it was one of those occasional (maybe quarterly) magazines like ATV
Magazine or something.

Again, the article didn't exactly say he had ever fired it up, so
reading between the lines seemed to indicate that it was a project in
process, so he may not have had all the details worked out. Still, a
scary power supply and project in general.

Google Groups gets some hits on "microwave oven transmitter",
especially in rec.radio.amateur.homebrew which discuss just this
subject.

Jeff Liebermann wrote:

> Let me know if you find the construction article. I do remember
> seeing (and ignoring) it about a year ago.
>
>
You provided this link in another forum.
http://lea.hamradio.si/~s57uuu/mischam/magnetr/

http://www.furriestdog.com/hipower.html

Bob