TWY, "I could only account for perhaps 1-2 amps of device off current in the current 2GHz P4 design based on device count/device width/device type and published device characteristics. Now granted, that may increase 3X or 4X at minimum channel length. But that will still be less than 8A."
So, it still does not explain the published
specifications of 17A. I remember suggesting that the
17A is the worst-case for mass-produced units, while
the published device characteristics are for cherry-picked
best samples. Is this totally out of line? Are there
other not-yet-discoverd (published?) sources for
such excessive standby current?
Please note that they call this problem as "leakage",
and it's severity was publicly acknowledged by Intel
on several levels. Of course I am well aware of the prime
purpose of that slide, but IMO it also slips an additional
quantitative information, that there are parts capable
of running at fairly higher frequencies, but they
suffer significantly from excessive leakage, which
obviously drives the parts out of thermal envelope, and
consequently from the market, so far.
"Of course, if a substantial amount of active current is included in that so called standby current, then the figure could be misleading."
I do not believe even for a second that there is any
active current involved, since projections from active
currents at nominal frequencies give the same answer,
17A.
- Ali |
