[Beowulf] 90nm opteron power usage

[Kozin, I (Igor)]

Mark, in your experience how big is the difference between 
one- and two-sided modules?

> > 22 watts for ONE dimm is significant. for a motherboard with 8 such
> > dimms you end up with a rather huge power budget.
> 
> hmm, I thought I was clear: the correct expected max dissipation is 
> about 5W per dimm.  the theoretical max for an unusual configuration,
> achievable only on a dimm-testing rig would ~22/dimm.
> 
> 
> > > > Anyone know how much the memory could contribute to the 
> overall power load?
> > > 
> > > fairly minor: I just looked at a micron 512MB dimm that had a 
> > > peak draw of ~8.5A at 2.6V -> 22W.  that's for a dimm with the max
> > > number of chips, operating bank interleaved, one read per cycle.
> > > a real system could never manage that.  a same-spec device with 
> > > half the number of chips dissipates half the power, and a more 
> > > realistic full-load dissipation is more like 2A -> 5W.
> > > 
> > > so figuring ~30W of memory in a dual-opteron is probably 
> reasonable.
> > > that's real dissipation, of course - call it 45W when 
> measured at the wall.
> > > 
> > > > > I managed to hook up my kill-a-watt to a dual opteron 
> 2.6 GHz, 4GB pc3200,
> > > > > and a 73GB scsi disk (a sun v20z).  The highest power 
> consumption I
> > > > > could find was was running 1 distributed.net client per cpu.
> > > > > 
> > > > > The results was 197-198 watts and 200-201 VA.  I'm 
> quite impressed,
> > > > > our 2.2GHz version of the same box is in the 270-300 
> watt range.