This actually doesn't have much effect in a properly fabbed chip using
current technology. It does play a major factor when a particle of dust
has interfered with the process. For example, if a bit of dust has
contaminated the mask, an outcropping can appear on a wire on the chip.
For example:
--------------------------------------------
this is wire 1
--------------------------------------------
------------^-------------------------------
this is wire 2
--------------------------------------------
Notice the little outcropping on wire 2. The wires do not touch
(yet.) The electric field at this point will be very high, and after
sustained use, it's possible that a failure will occur at this point.
I think that it's best to encounter these failures earlier rather than
later, when they are likely to be covered under warranty. Burn that
sucker in!
In another message in this thread, someone mentioned that processors
are always doing something and don't stop. This is not strictly true.
Under some OSes (namely those out of the Seattle area) the processor
is always polling for IO or NOP'ing or whatever. Under other OS'es
(namely those that end in NIX) when there's nothing to be done, the
processor is halted. A processor is then woken up by an interupt (usually
when an IO device finally finishes doing its thing). This does
change things like power consumption and heat generation--running
deschall could theoretically cause heat problems in a borderline
system.
-- Mike Gebis gebis@ecn.purdue.edu mgebis@eternal.net