Accuracy/reliability of "top" CPU usage numbers

[Cathey, Jim]

Again, I'm not speaking as to what Linux does, exactly.
But in general, things that are triggered on clocks and
that _end_ before the next clock interrupt tend to be
unnoticed when statistical sampling is used.  If, for
example, my clock-synchronized task consumes 90% of the
CPU time before the next clock tick, but then finishes
and goes idle, the sampler will never see it, and can
report that the machine is 100% idle, when in fact it
is only 10% idle.

Samplers work pretty well for tasks that run
longer than that.  If the tasks end up synchronized
with the sampler, that's when the lies can creep in.
(But it's not necessarily wrong, either.  The true
answer is: it depends.)

If a task is still running when the sample is taken,
then the accounting is a lot more accurate.

Clock-driven sampling is simple and highly portable,
and has been used in Unix since Day 1.  To use things
like TBR instead is a highly machine-dependent. An OS
like Linux, which is also targeted for machines that
do _not_ have a TBR or equivalent, probably does not
use two or more entirely different implementations
internally.  (But ask an expert.)

Yes, 'idle' is the balance of time.  Idle + used will
add up to 100%.

-- Jim