Elmer, <I didn't write ohm's law, go argue with him.>
Sure you didn't discover the Ohm or any other laws
due to apparent lack of analytical and thinking abilities.
The Ohm law is valid for homogeneous piece of conductive
media. A processor consumes the electric power
via switching of millions of gates under strict control
of the program it is running. It is hardly a simple
piece of conductor, and the effective processor
"conductivity" is a complex function of a
program. If it executes a HALT command as Windows
NT frequently does when idling, the processor
consumes almost no power. If it crunches frequently
alternating data pattens across FPU or alike,
it may consume power in some excess of average
power consumption.
If the max current ever happen in any real
program, it seldom happen for more than a few
microseconds out of the total run-time of the
application. Hence there is a term "typical"
power consumption, which is much less than the
"maximum current" times max voltage.
However, a CPU manufacturer MUST specify the
max.current because if the power supply fails
to provide this PEAK current for those rare
and short periods of time, the core voltage
may drop below the level of safe operation,
and the program may encounter a data error,
or the overall test may fail or an application
lock-up may occur.
All this whinning about "excessive consumption"
of 0.25um K6 processors is no more than nonsense,
and you know this perfectly. This subject was
discussed at length here when AMD introduced
K6-233 in 0.35um, and the consumed power was
even higher than the current 0.25um K6. Pentiums-II
at 300MHz on early 0.35 um process consumed
even more power, up to 40W on average, and the PC
industry has had some problems with overall heat
buil-up in some poorly ventillated cases. So what?
BTW, have you seen the size of this Xeon heat sink?
developer.intel.com
Now please shut up about this subject. |
