OT? power reqs of DDR2 chips

[Wolfgang S. Rupprecht]

Les <hlhowell at pacbell.net> writes:
>     For memory  and processors, the power factor increases with ram
> speed, 

Note: you don't want to use the term "power factor" here.  That term
is already has a very specific meaning when talking about power.  It
is the power adjustment you need to apply due to the phase angle
between voltage and current.  It is what makes measuring power
difficult to impossible if you only have a volt and current meter.

> so the access rate will definitely have impact on your loading,
> as will the ram path length (the traces act like a dynamic load until
> charged), so the longer the path, the higher the duty cycle for
> transition current.  compact boards are faster and more efficient, so
> you see lots of really tight layout on the fast blades, or really
> compact design in shuttle type cases to reduce path length and
> associated losses.

The disk throughput speeds are so slow that if that is the only access
going on that you are essentially idling the ram.  You will not be
seeing any significant power increase there.  Look at the ram's
datasheet numbers -- you get approximately a 10:1 power change for
using it hard.  Using it hard is something you'll only see during
well-coded block moves etc.  The power vs access count is essentially
linear with accesses and we are using maybe a few percent of the ram's
potential accesses.

-wolfgang
-- 
Wolfgang S. Rupprecht                http://www.wsrcc.com/wolfgang/
Hints for IPv6 on FC6 http://www.wsrcc.com/wolfgang/fedora/ipv6-tunnel.html