kill a watt?

Robert G. Brown rgb at
Fri Mar 14 08:51:31 PST 2003

On 13 Mar 2003, Jim Meyer wrote:

> On Sun, 2003-03-09 at 16:35, Jim Lux wrote:
> > Sure, it'd be nice if it had an external interface, but then it would cost
> > $500+, if only because it's a line connected device, and you'd have to
> > provide isolation, etc.  There's a web site out there where someone took it
> > apart... everything is floated at line voltage inside (which makes it muy
> > cheap) much like X-10 stuff.
> Heck, at the $45 price RGB noted, you should point a cheap webcam at it,
> take pictures at some reasonable interval, and write a script to OCR the
> pix and parse the output for monitoring.
> But maybe I'm just geeking for the fun of it. =]

I got the device yesterday, and it seems very lovely.  Not terribly
heavy-duty, but sufficient to monitor a single 15 amp circuit (and maybe
a 20 -- if their 1875 watt limit is average power).  I haven't used it
on a computer yet, but I learned that my vacuum cleaner draw about 1300
watts and my steam cleaner draws only 800-900 (guess what I was doing
yesterday evening:-).

I honestly think that the display is too boring to point a webcam at,
although it would be easy enough to do so.  I mean, you plug a load in,
cycle through the buttons, and that's it.  Sure, you work through state
a bit, but state doesn't vary THAT much on nodes -- pretty much unloaded
and a few states of load up to 100% CPU+network+disk all at once --
fully loaded.  Five minutes of monitoring and you should know pretty
much what a node will draw in its mostly likely state, 24 hours of
monitoring (so you can can get a decent loaded baseline) and you can
turn KW-hours plus hours (it measures both) into average draw.

After that, you're watching "nothing interesting".

I view it as more a tool to test individual systems for that critical
five minutes, or in a cluster/production test 24 hours, one time, and
then move on.  After that it would be less exciting than watching paint

I'll try to post some pictures of the unit in action later today, if I
make it in to Duke and remember my camera.  At last I'll be able to get
a real time read of the 2466 power draw...;-)


Robert G. Brown	             
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at

More information about the Beowulf mailing list