[Beowulf] power usage, Intel 5160 vs. AMD 2216

Jim Lux James.P.Lux at jpl.nasa.gov
Fri Jul 13 16:42:52 PDT 2007

At 02:28 PM 7/13/2007, Robert G. Brown wrote:
>>Tough.. the energy density of lead acid is really high.
>>Here's an example using one of those big 1 Farad 12V caps the auto 
>>sound people use.. say the UPS can take voltage drop of 2V on the 
>>"battery".  At 12V it stores 72 Joules. At 10V, you've recovered 
>>22J. That's about 1/10th second, assuming the UPS is 100% 
>>efficient, which it isn't.
>Ouch.  I have been schooled indeed, and of course I know 1/2 CV^2 and
>should have been able to figure this out myself.  Lazy boy.
>>The battery life *should* be a whole lot longer (Lead Acid 
>>batteries can have 10-20 year lives, as can NiCd), however, the run 
>>of the mill UPS doesn't treat the battery very well.
>I have never seen a NiCd last that long.  One is lucky to get a hundred
>power cycles out of them.

Something is seriously wrong.

Typical Lead Acid should take 1000 cycles (where a cycle is full discharge)

NiCds, properly charged and used, should last for tens of thousands 
of cycles (e.g. they use them in spacecraft orbiting the earth 14 times a day)

>NiMH aren't even doing too well in my copious
>supply of rechargable batteries at home.  And of course a car battery
>that makes it to seven or eight years is more the exception than the

That's more driven by exposure to high temperatures and vibration in 
the under-hood environment.

>>>especially at mass market prices.
>>That's the problem.. mass market means race to the bottom for 
>>quality and life, to reduce initial price.  You can get a PC UPS 
>>which provides 10 minutes or so at 200-300VA for $50-70 at the 
>>local big box store.  If you were willing to pay, say, $300-400, 
>>you could probably get 20 year life.
>I'd just like to get more than 1-2 years.  Laptop batteries too.  At
>three years there is a notable drop in charge time, no matter what you
>spent on them originally.

Just as with UPSes, they carefully design the battery charger/life to 
not be too long (hence too expensive).  It's a sufficiently well 
understood engineering exercise that they can draw a fairly accurate 
graph of charge capacity vs time (with the variability of mfr and use 
factored in).

you get real long life in a spacecraft application because they 
carefully hand select and match the cells, and very carefully manage 
the charge and discharge profiles.

FWIW, the Mars Rovers use Lithium Ion batteries, with nominally 1000 
cycles life. (which they've exceeded by now) 

More information about the Beowulf mailing list