Re: [stella] And now, for my next trick...

Subject: Re: [stella] And now, for my next trick...
From: "Andrew Davie" <adavie@xxxxxxxxxxxxx>
Date: Fri, 9 Mar 2001 01:12:31 +1100
> > b) Life on Mars.   An implementation of John Conway's life (in at least
> You ain't the first attemtping that:
> - Check in the archive for "life;game" to see what happened to the
> attempts of John V. Matthews and theories about it.
> - Check espacially for a thread called "Atari Life" with additional
> info.

Thanks for the pointers.  Nobody seems to have considered packing the data
in RAM.  Considering that MOST of the playfield during a life 'game' is 0
(the theoretical maximum is 50% used), and that large areas of the playfield
are usually contiguous 0's... and also that non-zero and alternating
patterns are typically grouped together...  I figure a simple run-length
packing (speed essential!) would be feasible.

All the messages I read seem to assume that 1 bit/pixel is the limit that
can be achieved.  This is not so!  As a simple example, byte-aligned runs of
0's only (eight 0's) could be efficiently packed to a single bit, and all
other data left as-is (plus another bit).  This would be pretty quick to
unpack, yet save HEAPS of RAM.

I'm still thinking about the feasibility and operation of the kernal, and
how I can interleave run-length unpacking into the display kernal, in
particular.  I'm not saying it CAN be done... I just think it might be
possible :)

 _  _  _| _ _        _| _    * _                               _  ,
(_|| )(_|( (/_\/\/  (_|(_|\/(_(/_                           ,~' L_|\
                                                         ,-'        \
see my Museum of Soviet Calculators at                  (            \              \    __     /
                                                          L,~'  "\__/
                                                              @--> v

Archives (includes files) at
Unsub & more at

Current Thread