Re: 6702 chip

From: Segher Boessenkool <segher_at_kernel.crashing.org>
Date: Fri, 11 May 2012 19:48:27 +0200
Message-Id: <E75D747E-C6D4-4119-BE2E-4AA6F0ECA4F9@kernel.crashing.org>
WARNING SPOILERS
DO NOT SCROLL DOWN IF YOU LIKE TO PLAY




On Fri 11 May 2012 at 09:58:58 +0200, Rhialto wrote:
> For the moment I have pasted my lists into pastebin:
>
> http://pastebin.com/c2JNM5DA
> http://pastebin.com/mBjqdb8k

Thanks, those clarified things :-)

[skipping some 80 lines or so...  spoilers ahead...]









































































Still here?   Okay.

Each bit in the output is constructed independently.


[skipping some 80 lines or so...  more spoilers ahead...]







































































Every bit is constructed via some ring counter.  Not exactly
LFSR, but close.


[skipping some 80 lines or so...  final spoilers ahead...]







































































You should include the first $d6 value in the table as well,
it make it more regular.  With that done, if you look at only
at one, it stays the same for N cycles, then flips, and the
other level for N cycles.  For some of the inputs it does never
flip, stays constant all the time.  For bit K in the output,
only bit K in the input matters.  The half periods for the bits
are, highest bit to lowest: 2 5 3 1 8 7 3 6.

[okay then, final final spoilers ahead]





































































The circuit for each bit is a shift register with feedback, and
the value you input is also put into that feedback.  To find out
the exact feedback formulas, you'll have to put some non-constant
patterns into the chip.

This is almost a set of LFSRs by the way; except the feedback is
not linear.  But similar analysis techniques apply.


Segher


       Message was sent through the cbm-hackers mailing list
Received on 2012-05-11 18:00:30

Archive generated by hypermail 2.2.0.