On Mon, 28 Feb 2000, Shawn T. Rutledge wrote:
Not sure I've ever seen one... can you put up some
pictures?
Yes, I had planned to. My SLR has been cranky recently, and I needed some
photos from an event I performed at so I picked up one of those single-use
jobs and have been filling up the rest of the roll with stuff from my
collection. Let's hope they turn out respectably. (:
One of the things I wanted to accomplish is to illustrate the amazing
amounts of miniaturization 13 years can see. For example, one of the
photos will be of one of the 1 MW ECC memory boards next to a pair of 32
MB ECC SIMMs from an HP 715.
Anyway.
I heard they had nice keyboards with unshifted keys
for all the bracket
variations (<[{) so you don't have to keep hitting shifted-numbers all
the time in lisp.
The parens are where they normally are (over 9 and 0), but are also
available unshifted where the square brackets would normally be. The
square brackets have been moved to where the curly brackets would normally
be (on this keyboar, above the unshifted parens) and the curly brackets
have been moved to shifted keys which are in Siberia.
What I find most amusing is the sheer number of 'modifier' keys on the
keyboard. Caps-lock, shift, symbol, hyper, super, meta, control, and
mode-lock. Plus perhaps one or two more which work as modifiers but which
I don't recognise as such.
According to the docs, the keyboard has 'unlimited rollover', meaning if
you hit a key, the machine takes note of it no matter how many other keys
you're holding down at the time. Mind-boggling. Apparently someone saw a
great need for a user to be able to type an n-bit character set directly
from the keyboard, where n is some ridiculous number
like 12.
Didn't know they used a GUI though. Is it X
window or something special?
I'm not sure, but I don't think that X11 existed at the time this machine
was current in 1987. X10, maybe, but the windowing system is its own
thing. Nothing too fancy. I'd put it on par with the Apollo DM in terms of
the kind of functionality it attempts to provide.
Do they still have performance advantages over less
specialized systems?
What kind of hardware modifications would optimize execution of Lisp?
Still? I'd be surprised. According to the User's Guide to Symbolics
Computers:
"The power, speed, and flexibility of Symbolics processing machines result
from optimizing the hardware design to match the
software environment.
Some of the special architecture features include:
Tagged architecture
Multiple caches
Hardware stack management
Pipelined instruction cycles
Parallel processing
Hardware assisted garbage collection
Fully ECC'ed system memory
(end of quote)
I'm very intrigued by what I've learned about the system so far and am
anxious to start it up and give it a try.
I get the feeling that the LISPM is something of a transputer (like the
parallel i860 boards that Microway used to sell which fit in an Intel PC),
but with the balance of power between the transputer board and its host
(LISPM and FEP, respectively) swapped around so that instead of the PC's
OS hosting software which crunches away on the transputer, the OS actually
runs on the LISPM, and the FEP's job is to boot the environment and then
to handle I/O once the environment is running.
I don't feel like I've described it very articulately, but I only vaguely
understand what's going on at this point anyway. Still. It's an
interesting conceptual leap!
ok
r.