From: Humphreys, David (david.humphreys_at_honeywell.com)
Date: Wed Jun 27 2001 - 07:01:30 PDT
Hi to all you overclockers,
I have been following the thread on overclocking
and would like to give my opinion on the subject.
As some of you have no doubt found out, it is possible
to speed up the MP2x00 by swapping the clock crystal.
It has been advocated by some that all that is required
is a switch and two crystals.
As a designer, I have to warn you that, while this
method might work, you run the risk of corrupting your
data.
Here's why:
Changing the clock frequency with a switch is asynchronous
that is to say, the action is not tied to the clock cycles.
The switching, therefore, takes place at a random time
during the clock cycle.
The memory, asic and processor need to complete their respective
tasks within a set number of clock cycles and are synchronous
to them.
Potentially, switching could cut short a clock cycle or cause
a glitch on the clock line that disrupts an internal task.
What this means is that you could get, among other things,
corrupted memory.
The effect, depending on position, may be very subtle and
go unnoticed for some time, perhaps corrupting more and more
data.
And for those of you thinking that you could just turn off
the unit, switch the crystal and then switch back on, let me
tell you that the Newt never sleeps it just reverts to a low
power mode.
The Pix solution board has the additional chips on board
that take care of the switching. It will wait for a clock
transition before switching.
The last part of the puzzle is the SA-110S. The part is
stamped 160Mhz. There is a reason for that.
A lot of chips have frequency ratings. When they are made,
the manufacturer can't afford (or doesn't want to) different
production lines for different frequencies of the same
chip.
Instead, they will make the fastest chip and test them.
Those that fail at, say 220Mhz may pass at 180Mhz or 160Mhz.
The point I am trying to make is this:
The processor started out life as a 220Mhz device but factory
testing found it to fail at this frequency and it was tested
at the other speeds until it passed. Then it got stamped with
that number.
Manufacturers tend to err on the side of caution, so it is likely
that parts will work 1 or 2 speed grades above advertised
but YMMV.
Also, higher speed means higher power dissipation. This will
prematurely age the processor and other components.
I know the lure of a quick and easy and cheap solution for
a faster Newt is tempting but it is not without risks.
I am not telling anyone not to do this, I just wanted you
to make a more informed decision.
I would be happy to share my knowledge with anyone who asks
for it.
If you have a question, drop me a line
mailto:pcbman_at_ix.netcom.com
or
mailto:david.humphreys_at_honeywell.com
Regards and sorry for the length,
David Humphreys
Designer
Honeywell Inc.
-- This is the Newtontalk mailinglist - http://www.newtontalk.net To unsubscribe or manage: visit the above link or mailto:newtontalk-request_at_newtontalk.net?Subject=unsubscribe
This archive was generated by hypermail 2.1.5 : Sat Mar 08 2003 - 11:37:04 PST