DIYbanter - View Single Post

Doug Miller

In article , (Dave Martindale) wrote:
(Doug Miller) writes:

Whether it's a 1/4 mile drag strip or the Indy 500, the top speed of my '57
VW Bug is the same. The length of the track or, as you put it, the signal
path, has nothing to do with the speed.

Length of path x speed = time required to traverse the path.
Since the propagation speed of electrical signals has an inherent physical
upper limit, the length of the signal path places an upper limit on the speed
of any device that is depending on those signals.

Actually, he does have a point. Transmission delays through wires and
other bits of hardware set a floor on latency - the time it takes for an
unpredictable input to influence the output. But it doesn't limit
throughput. In the same way that a CPU with a long execution pipeline
may have many computations "in process" at a given time, communication
links may have several signal states "in flight" within a single cable
at a given time.

Yes, and if the processor clock rate is high enough, and the cable long
enough, the processor can wind up having to wait for a signal to arrive --
which *does* limit throughput.

--
Regards,
Doug Miller (alphageek at milmac dot com)

It's time to throw all their damned tea in the harbor again.