Hi,
Dmitry wrote:
>Nultiplying by constant in such way what every original bit from A
>being transferred into reversed position in result's high register.
>Since multiple positions resulted divison is used at the end.
>Can be implemented on PIC's with hardware multiplication as
>well I think. I like the way Gosper and Freiberg thought. ;)
>
>Add/sub operations are capable to shift bits simultaneously to
>the same distance while mul/div can move each bit differently.
>Interesting...

Yes interesting indeed, last time this came up I spent a few minutes
trying to use the multiplier on the 18xx chips, I only got 'halfway'
though. In about 6 cycles (I think?) I got the following:

Input: b7b6b5b4b3b2b1b0
..
6 cycles :)
..
Output: b1b0b3b2b5b4b7b6

But it had me stumped there, lacking time (and brainpower :) ) I had
to leave it at that state. Of cource it's 'easy' to correct it but
the number of cycles it took seemed ineffective, I feel there has to be =
an
better way (gut feeling).=20
I think with the limited multiplier (8x8) it's hard to reach an =
substancial
improvement over the 'brute force' method (shift->carry).
Anyway this is a brain teaser indeed.

/Tony

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics