The Amiga's graphics blitter chip was smart enough
to perform
the operations of Life. According to some messages I found,
on 7 Mhz 68000 machines, it would do 1.3 million cells/second.
Someone else reported 19.8 generations/second for a 318 x 188 grid.
I ran that, ages ago. Boy was that slick. The original is on an
older Fish disk (c. 1987). A slightly updated version (c. 1994)
that descends from that one is available at
http://wuarchive.wustl.edu/pub/aminet/game/misc/fastlife-2.7.lha
I'd give a URL for the 11-year-old version if I could, but I
don't remember the FF disk number.
On an unexpanded Amiga 1000, Tom Rokicki's hardware assisted version
of Life would run so fast that it looked like a time-lapse of bacteria
growing under a microscope. I had run versions of Life on the VAX at
work (11/750 w/8Mb RAM, <20 users, VT100 @ 9600) and seen a generation
every couple of seconds. It's especially telling because the underlying
processor, the 11/750 is rated at .6 MIPS (VUPS), about the same
as an 8Mhz 68000. The difference between a 318x188 grid at 20Hz
and 80x25 at .5 to 1Hz is a) the blitter and b) the serial port
speed, but to see one go, then the other, the change is several orders
of magnitude (displaying 1.2x10e6 vs 1.2x10e1 cells per second). I
suspect that if you had a framebuffer on the VAX, you could get
within 1% of the Amiga's speed.
-ethan