On 05/13/2011 04:33 PM, Ethan Dicks wrote:
On Wed, May 11, 2011 at 1:23 PM, Pete Turnbull
<peter.turnbull at york.ac.uk> wrote:
No, but very cool. I'm not suggesting
uber-tight packing, but it'd be
interesting to see if that could be tweaked to do maybe 8 or 16 bytes
- that should still be loose enough to run successfully by hand.
-ethan
Many years back I did a two wire core plane for 16 bytes.
I used cores just slightly larger than those in the early PDP8
that some one gave me a load of. To go larger than maybe
64words/bytes it's better to do a three or four wire approach,
two half selects and a sense/inhibit line or separate sense and
inhibit. I didn't go there as the finest wire I could find was too
thick for more than two wires (select and sense).
The hardest part was finding the "write current" and Slice
or read timing. next step was write and inhibit.
One challanege was sensing the valid read after the huge
select pulse. it's small compared to the induced pulse
from the select current.
What starts to happen as the size of the matrix grows
the number of external support parts goes up till you hit
some large number like a few K and 4/3 wire geometries.
Also you need 1 bit per bit of data (8 for byte 16 for
16bit for example) to hold the captured data during write or
read. most system use core as RMW cycle, that is a
destructive read, execute, modify if required and then
write back.
Allison