[Beowulf] /. Cooler room or cooler servers?

Jim Lux James.P.Lux at jpl.nasa.gov
Thu Apr 7 19:11:41 EDT 2005




At 03:29 PM 4/7/2005, Mark Hahn wrote:
> > No mention of PowerPC which runs far cooler (1/2?) than Intel/AMD:
>
>except that it doesn't.
>
> > http://www.terrasoftsolutions.com/products/apple/power-useage.pdf
>
>lovely graph.  alas, the numbers are pure fantasy.  yes, I actually
>did measure a dual xserve g5 (not heavily configured) and it was 200-210W,
>quite comparable to dual-opterons and even dual-xeons.


Leaving aside such tantalizing issues as what is a GHz, in terms of useful 
work.

However, if we do a straight watts/GHz, Xeon is 34.375, G4 is 33.8 ... not 
too much difference.

One also needs to look at important ancillary issues like power consumption 
of the cache, bus drivers, etc.  And, of course, instruction stream makes a 
big difference.

I have a DSP system down in the lab that happens to be instrumented for 
these sorts of things.  The power consumption while running a program 
varies by a factor of 10:1.  During FFT's the power consumption is very 
high (since both ALUs, and all the addressing registers, and pipeline 
registers are running at peak performance).  When moving data off chip to 
external devices, the power drops way down.  You can see the end of each of 
the 9 passes in the FFT, as a dip in power while the indexing is 
reinitialized, etc.

In fact, the ability to see the instantaneous power was an extremely 
valuable debugging tool on a processor where we didn't have a lot of real 
time visibility.  An interrupt coming in at the "wrong" time, or an 
inopportune context switch, was painfully obvious on an oscilloscope.



James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
tel: (818)354-2075
fax: (818)393-6875


_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf



More information about the Beowulf mailing list