🚀 go-pugleaf

 bob@reed.edu (Robert Ankeney) writes:
>     One thing I was just remembering was a wonderful puzzle from way
>back when.  It was a challenge to write a program that zeroed memory.
>It had to be written in the first 4K and would zero the entire 4K,
>including itself.  So, if you're feeling bored, you might see if you
>can do it!  Sadly, I don't remember any solution, but if I recall,
>the auto-increment registers at 10-17 might have been useful.

A puzzle?  We ran this little ditty on almost a daily basis.
0000 ADDR LD
DCA I 0010
JMP 0000
0000 ADDR LD CLEAR CONT

2 instructions.  Actually, we'd run it with the data field set to 1,
then 2, and finally zero.  The program autoincrements itself right
through the field, finally (in DF0) overwriting the DCA with a zero.
(AND 0000)  Then the loop was a lot tighter -- you could see the
difference in the panel lights.  You do have to make sure that
location 10 has a 10 in it when you start, though...

Why did this get run so often?  Because the teachers didn't know how
to reload BASIC, so if the machine went down they'd have to corral a
convenient student, write him a pass out of his next class, and "make"
him reload it.  If you didn't have a panel key, this program would
bring EDUSYSTEM-20 to its knees:
10 DEF FNA(X)=FNB(X)
20 DEF FNB(X)=FNA(X)
30 PRI FNA(1)
RUN
<CRUNCH>

Another interesting pdp8 trick:  Since we had (4) ASR-33s as our only
IO, we wrote a "fourple loader" which would read a clunk from each TTY
in turn.  (The first character was on tape 1, the second on tape 2,
etc.)  This ended up being almost 4 times as fast as loading
through 1 TTY, since most of the time was spent just waiting for the
reader to come ready again.  Each TTY used a tape with a different
leader character, and the loader would figure out which tape was in
which TTY.  Nowadays I think they call this "reverse multiplexing."

-- 
----------------------------------------------------
Gordon S. Hlavenka           cgordon@vpnet.chi.il.us
Vote straight ticket Procrastination party Dec. 3rd!

In article <1992Nov19.102634.20745@vpnet.chi.il.us> cgordon@vpnet.chi.il.us (gordon hlavenka) writes:
>Another interesting pdp8 trick:  Since we had (4) ASR-33s as our only
>IO, we wrote a "fourple loader" which would read a clunk from each TTY
>in turn.  (The first character was on tape 1, the second on tape 2,
>etc.)  This ended up being almost 4 times as fast as loading
>through 1 TTY, since most of the time was spent just waiting for the
>reader to come ready again.  Each TTY used a tape with a different
>leader character, and the loader would figure out which tape was in
>which TTY.  Nowadays I think they call this "reverse multiplexing."

I love it!  Paper tape striping!

Jeez - if you could add enough serial ports, you could attempt paper
tape virtual memory.....    nah.
-- 
Jonathan E. Quist                              INTERACTIVE Systems Corporation
jeq@i88.isc.com  '71 CL450-K4 "Gleep", DoD #094        Naperville, IL
        There are many things a person should experience in a lifetime.
       Among them are an infant's first cry, and an infant's first laugh.

In article <1992Nov19.102634.20745@vpnet.chi.il.us> cgordon@vpnet.chi.il.us (gordon hlavenka) writes:

>
>A puzzle?  We ran this little ditty on almost a daily basis.
>0000 ADDR LD
>DCA I 0010
>JMP 0000
>0000 ADDR LD CLEAR CONT
>
>2 instructions.  Actually, we'd run it with the data field set to 1,
>then 2, and finally zero.  The program autoincrements itself right
>through the field, finally (in DF0) overwriting the DCA with a zero.
>(AND 0000)  Then the loop was a lot tighter -- you could see the
>difference in the panel lights.  You do have to make sure that
>location 10 has a 10 in it when you start, though...

When the DF is 10-70, it doesn't much matter what's in 00010, since you just
let it run more than enough times to clobber everything many times over.

With the DF 00, when the DCA I 10 is destroyed, the JMP isn't destroyed.
However, since the switches are zero from the last load address function, the
switches are thus conveniently set to 0000.  Hitting deposit will put a 0000
into the latest location, which since the running program is now:

	*0000

	AND	0
	JMP	0

means that the PC is either 00000 or 00001, so worst case you can either
deposit a 0000 over 00000 and 00001 if you hit deposit twice, or a 0000 over
00001 and 00002 if you hit deposit twice.  This can be done without looking at
the lights if you're in a hurry.  An alternative is to do ADDR LOAD then
DEP DEP but that might take most people a few extra milliseconds.

Here's one for the fans of the (nearly) cleared-out memory PDP-8:

Assume that all of field 0 is all 0000 as the above might accomplish.  ADDR
LOAD 0000, change the switches to 1777, press DEP, CONT.  Describe what the
AC lights are doing, and why.

(On a difficulty scale of 1-10, this problem is rated 2.)

cjl

 lasner@watsun.cc.columbia.edu (Charles Lasner) writes:
>Assume that all of field 0 is all 0000 as the above might accomplish.  ADDR
>LOAD 0000, change the switches to 1777, press DEP, CONT.  Describe what the
>AC lights are doing, and why.

That's a weird one!  I just tried it, and now I'll have to sit down
and cipher what is going on in there...

>(On a difficulty scale of 1-10, this problem is rated 2.)

This means you have more?

-- 
----------------------------------------------------
Gordon S. Hlavenka           cgordon@vpnet.chi.il.us
Vote straight ticket Procrastination party Dec. 3rd!