In article <R_6%b.22243$H17.6738@twister.nyc.rr.com>,
sweet <sweet430@hotmail.com> wrote:

So I am asking: how is it that the same wu can return different results from
different machines.

You cannot draw that inference from what I wrote.  Agreement only has to
be within the expected computation errors.  The cross check is to eliminate
the results of machine faults (e.g. from exceeding the machine's tolerance
to running with the clock out of specification) and people who cheat and
don't actually compute the unit.

                    Is the crunching algorithm that different for each type
of machine?

The two main causes of difference are:

- the chirping calculation is done incrementally and occasionally recalibrated;
  however, if you stop and start the client, it is done accurately for the
  current chirp rate; this means the accumulated rounding errors can differ;
- different compilers will use machine registers in different ways; machine
  registers generally have more significant bits than even the largest
  format of number in memory - most calculations are only stored in single
  precision.

Please do not start a new thread for what is a continuation of an older
thread.  Also, it is unwise to personalise subjects by using correspondent's
names, and, in any case, your subject is not helpful to anyone trying to
quickly select which threads to read.

A correctly posted followup should include a References header listing
the message IDs from the start of the thread up to the the replied to
article.

Thread broken from: <T1077695816@djwhome.demon.co.uk>
Thread root: <6DJZb.214795$4F2.29896791@twister.nyc.rr.com>