Conference noted::hackers_v1

    
    When I was back in High school they had a PDP-8/E system.
    
    Then about 1 yr. ago they were getting rid of it.
    
    So now its mine (all 3 racks worth)
    
    

    Well, I remember first using an 8/L in college.  After having worked
    with the IBM beastie (7044, would you believe!), one of the first
    "programs" I wrote for it was about 5 instructions toggled in from
    the switches.  When I ran it, it printed an "S" on the TTY!  At
    that point, I had the revelation that *now* I understood how computers
    worked!

        Hmmm...
        
        I've never have the opportuniity to use a PDP-8.  When I started
        college (1981), Northwestern had just gotten rid of its PDP-8's
        for the 'modern' PDP-11s.  However, our Computer Subsystem class 
        had us (as a term project) design (a simplified) PDP-8 control 
        unit.
        
        BTW, I once heard that there was a wire screwed onto the back
        of the PDP-1.  It did not lead anywhere; however, when removed,
        the machine would refuse to run.  Is this true?

       -Mike

    [RE .2]:  I wrote a 5 instruction program that would wipe
out 4K of memory . . .
		<_Jym_>

Ah, nostalgia ain't what it used to be!

My first summer job was to write a data transfer program for a PDP-8
in a parapsychology lab to transfer data from a twelve track analogue
recorder (raw sensor data) into some meaningful form on paper tape
so it could be loaded onto the mainframe (an old PDP-10).

No problem -- I'd written programs in University to use the -8 (hell, I
think I've still got the RIM loader written somewhere), so off I went to
start work. They showed my my office, where the coffee pot was, and then took 
me into the lab to show me this "PDP-8".  Ever seen a LINC-8? The "sort-of"
dual processor -8 with left and right registers, etc etc etc? Surprise, Nigel!

Well, that was an interesting summer. 
	NAC

    
    	And, there are people still using PDP-8s. In a DuPont computer
    room I went in, they had 2 PDP-8s, 2 KA-10s, a 10 node VAXcluster,
    an IBM 43-something and a CRAY-1 ! Quite a mixture I thought...

    
    Trev

    Re .6: Were they using the 10 node VAXcluster as a front end to
    the Cray? :-) :-) :-) :-) :-).
    
    Mike

    [RE .4]:  That 5 instruction program to erase 4K core is in the
    
               PDP-8 maint manual.
    
    
    

    re .4:  One of the technological improvements with 11's is you can
    now wipe core with one instruction - MOV	-(PC),-(PC)
    
    -Madman

    Ahh yes, the PDP 8!  My random memories.....

    	This is the best i could come with for clearing memory.  I think there
    was a 2 or 3 location rtn to do it.
        
	* 7774	7600 cla 
	* 7775	3010 dca 10
	* 7776	3410 dca I 10
	* 7777	5776 jmp .-1
    
* Remember the hack in COS-300 (I think Richie Larry wrote it) That
    determined if the CPU was a Digital PDP8 or something else.  If
    it was something else, I believe the disk bootstrap was changed
    from a read to a write.
    
* A realy nice thing about the 8's I worked on, was I was never afraid
    to reach over and press the Halt switch.
    
* Remember sweating for HOURS and HOURS for some way to get a free location
    on a 177 location page?  Finally I would do a JMP indirect using
    some instruction as another page pointer.
    
    Thanks for the opportunity for a former 8 hacker to trip down memory
    lane.  tim...

    Your routine does not clear out location 7777.
    Also, pages were 200 (octal) words long, not 177.

The first computer I recall using was a smallish PDP-8 connected to
an ASR33 which was running a Tic-Tac-Toe program which learned.  It
was at an open house at Brookhaven National Labs.  Years later they
had the same -8 and terminal running the same program in the Brookhaven
Graphite Research Reactor (BGRR), which is a big cube that used to be
a reactor, but is now the visitors center.  They also have a broken
looking pair of mechanical waldos that you can play "put the bolt into
the hole" with and a machine that measures radiation contamination on
your hands and feet and displays results with neon decade counters.

The summer before college I almost got a job in the chemistry building
hacking PDP-11 system hacks for a PhD that was into displaying molecules
in 3D by crossing your eyes.
				/AHM

    [RE .12]
    
    On a PDP-8/E 
    			Memory pages are numbered, in octal, from
    			page 0 (0000 to 0177) to page 37 (7600 to 7777)
    
    

    	I first started programming using a PDP8. We were doing speech
    processing. There was a DMA link to an IBM 360-30, and we used that
    as a backing store. It was faster to do the processing on the PDP8
    since that had 12 bit data paths while the IBM was only 8 bit. It
    made quite a bit of difference since our A/D converter was 12 bit
    samples.
    
    	Later on we got the money for our own disks, but it turned out
    that it was cheaper to buy a PDP8/E with disks than to upgrade a
    PDP8 to take disks. However, since all our old peripherals (including
    the vital A/D and D/A converters) were the PDP8 -ve bus they could
    not be moved to the new machine. So I had to design, build and
    programme a link between the two machines.
    
    	The most interesting hack was the binary loader, that fitted
    in the conventional space (7600-7777), but would accept an OS/8
    binary file PIPed to the link, complete with handling the link error
    detecting protocol I had to put in. 
    
    It was nice working on a machine which you could fix by replacing a 
    transistor. Just try that on a broken 8600 :-)

    My first "PDP-8" was a PDP-5. Transistor based machine (no chips).
    Interesting system; there was no hardware program counter; location
    zero was the PC; it was fetched in autoincrement mode and then used.
    This allowed you to do neat calculated jumps. (i.e. DCA 0 and away
    you went).
    
    Anyone else remember TSS-8 (aka EDUsystem-50?) 24 users on a 32K
    PDP-8, running it's brains out. Or P?S Checkers on the VT8E? or...
    
    Some of my most intense hacking was done on the 8's, trying to bum
    that last location off a page. Instructions as data, and so forth.
    My favorite quote, Elekman's Theorem (hope I spelled it right) went
    something like:
    "With the proper effort, 201 words of PDP-8 instructions can be
    fit onto a 200-word page"
    And a Richey Lary comment:
    "The difficulty of applying Elekman's Theorem varies directly with
    the number of times it is applied"
    	-Rick
    	[Author of WPFLOP, RTFLOP, OS/8 Adventure...]

    [RE .9 [RE .4]]:  You had a maintenance manual?  We used to
dream about having a maintenance manual!  We had to read the
holes in the paper tape to find out how the BIN Loader worked
(before I wrote some dump programs, that is - one that dumped
the tape and one that fit in the BIN Loader's space and dumped
the memory).

    [RE .16]:  Those are good theorems.  I learned quite a few
nasty hacky tricks, which have a permanent influence on my
programming style.  Who would have thought that a book with
the innocuous title "Introduction To Programming" (I believed
it - my first language was BASIC and my second was PAL-III)
would be an introduction to arcane hackery?

    [Another Anecdote]:  The BASIC interpreter we had was a
package of two tapes.  The first was the interpreter itself.
The second was a "reconfigurator" or somesuch.  If your compiler
stopped working, one loaded in the reconfigurator, which would
repair areas of memory that DEC felt were likely to be screwed
up.  This usually worked and was faster than loading the entire
interpreter (remember, these were *paper* tapes).
    I was getting on the airplane, on my way to become an ex-
change student in Germany, when the head of the math department
showed up, a little out of breath, with a packet of paper tapes
in his hand.  The BASIC interpreter tape had broken, and they
couldn't get BASIC running, and what should they do?
    And so, as we shuffled our way onto the plane, I looked
over the two BASIC tapes.  They both had the same BIN Loader.
The BASIC tape had, fortunately, broken in the middle of the
BIN Loader.
    I took out my green pen and drew a line on each tape, right
after the last instruction of the BIN Loader.  Then I explained
to him that he should load in the reconfigurator tape and stop
it right before the green line.  Then he should take the BASIC
tape and position it so that it was right before the green line
(the same number of holes, of course), and load the rest of the
tape in.
    This is how they kept BASIC afloat until I got back from
Germany and made them a new tape.
		<_Jym_>

    
    
    [RE .16]
    
    I have TSS-8.
    Would like to use it but can't!
    
    RF08  CRASHED!!!!! Piece of @%&$!   
    
    
    
    

    	I never used TSS-8, but I do remember its date problem. It was
    one of the first problems I was asked to fix when I joined Software
    Support.
    
    	I never used a PDP-5 either, but I do remember the hack in Focal
    that would allow it to take an interrupt through either location
    1 or location 0 depending on whether it was running on a PDP-5 or
    a PDP-8
    
    	Now how many remember CAPS-8 or ODP-143?  I had to make these
    two work together, and since they both expected exclusive control
    of the PDP-8 interrupts it was not too easy.
    
    	Hacking ain't what it used to be! Its too easy in DCL.
    
    		Dave

    Re: .18:
    I had (and may still have somewhere) a version of TSS-8 that used
    an RK05 for the disk. Actually, it used the A "side" of the pack
    (inner tracks), so you could run OS/8 on RKB0 and TSS-8 on RKA0.
    Not quite as fast as a RF08 swap disk, but it was a *very* good
    system-level diagnostic as well as a drive exerciser.

    TSS-8 was an interesting product. I was responsible for
    supporting/maintaining it when I worked in Corporate Field Service.
    Not exactly the normal organization for supporting software products.
    
    Re: .19:
    Yup, I remember CAPS-8. What was ODP-143, though?
	-Rick

    	ODP-143 was a symbolic debugger for the PDP-14. It was designed
    to run on a PDP-8 host under OS/8.  Under OS/8 there were no problems,
    since that operating system did not use interrupts, and in fact
    left the interrupt address free for application software.
    
    	I hear we have recently re-invented the same sort of thing for
    debugging VAX ELN programmes  :-)
    
    		Dave
    (P.S. I still have my PDP-14 reference card, just in case)

    Do anyone out there have any stuff around for the PDP-8/(E or A).

    I still have copies of some games as well. ANIMAL amongst them.
    (the worlds second AI program after LIZA).
    And a neat SPACEINVADERS program to run on DECMATEs (=pdp8)
    
    I am cheating though, I keep them on 8 inch floppies, not papertape.
    
    And then I have this card deck with an ship track simulator
    program for a xerox-7 beast.
    

    I've thought that a neat hack would be to build a PDP-8 emulator
    for the VAX on the lines of the PDP-11 emulator for non-compatibility
    mode machines -- you build a table of routines for every opcode
    and call through it.
    
    	Jon

    There is a PDP-8 emulator which runs on the PDP-11/60 under RSTS/E.
    It uses the writable control store feature of the 11/60 (there is
    an undocumented directive to write the control store).
    
    The WPS group uses that for their development.
    
    Kevin

  I know there where many 8 systems made.

  Ones that I know of that were hardware compatible are (OMNIBUS):

   PDP-8/E      QUAD
        /F      QUAD
        /M      QUAD
        /A      HEX


  What other ones where there?

	My first job involved writing an extended OS-8 assembly language 
interpreter for the 11.  We later moved it to the VAX.  I decided to leave that 
company about the time they decided to rewrite all the OS-8 code in a lower 
level language - COBOL.

Steve

    NOSTALGIA????!!!
    
    Us over here in DECmate land are STILL using the PDP-8... 
    
    The DECmate IS a pdp-8 (Harris 6120 chip)...
    
    It's not dead yet...
    
    
    Don Mac

    So maybe there IS a PDP-8 in there; but WHERE'S THE LIGHTS?
    Where's the SWITCHES??
    
    Getting so's you can't hardly tell something's a computer these
    days...
    						- Bluejay Adametz, CFII

    	In a *REAL* computer you can replace a transistor (or valve)
    in the accumulator. DECmate must be just a virtual computer. :-)
    
    	Now if you really want to be nostalgic (or thrive on trivia
    questions), how many of you could describe a decatron?
    
    (As a clue, it is a computer component, bue even I am not old enough
    to have actually used one).
    
    		Dave
    

    
    I used a PDP-8/E back when I was in high school, um, er, something
    like 1977-78.  They never did get rid of that thing, and last I
    knew, they still had it.  Its only problem was it took 25 minutes
    to compile an OS8-BASIC Star Trek game, and we usually would end
    up burning out a DECtape drive, once every other week, or so. :^)
    Never in my right mind, had I known I would be working for the same
    company that manufactured it.  Also did some front end programming
    with the switch panel.  Yep !  Those were the days when I didn't
    know what I was getting myself into, and considered using a computer
    fun !  Now, eight years later, I sit with blood shot eye's, shaking
    hands, bags under my eyes, and think the IBMer's are the _real_
    comunistic threat to the nation, and wonder if any of this was a
    good idea. :^)

    But then again, I wouldn't have it any other way ! :^)
    
    Matt :^)

    We also had a PDP-8/E in high school.  Complete with four teletypes,
    an RK05 and a DECtape.  We were all thrilled when the LA36 showed
    up.  'Look at this printer fly!  A whole 300 baud!'
    
    Last time I went back, the PDP-8/E was replaced with an 11/34 running
    RSTS/E.  You call that moving up?
    
    tom_p

    I just remembered a wonderful Stan hack of OS/8.
    
    The OS/8 CCL commands were designed to look somewhat like tops-10.
    i.e. EXECUTE foo, LOAD foo, etc.
    There was a nifty variation to SET TTY - SET TTY GAG.
    It printed out a (rather horrid) joke.
    	-Rick

	    What was the horrid joke?
			<_Jym_>

    Well, not having a system that I can run it on. (I'm at home) it
    went something like:
    
    A beggar approached me on the street.
    "Sir, do you have $100 for a cup of coffee?"
    "What! $100!" I asked.
    "Well, ", he said "they've converted everything to computers,
    and processing the form costs $100!"

    It's not quite right, but it's close enough.
    	-Rick

(For nostalgic reasons, the remainder of this note will be in upper case)

MY FIRST EXPERIENCE WITH A COMPUTER WAS IN 7TH GRADE ON A PDP-8/E RUNNING
EDU-25 BASIC.  I WAS EXTREMELY FOND OF PLAYING COMPUTER GAMES AND SOON
STARTED WRITING MY OWN.  ONE OF MY PROGRAMS WOULD ALWAYS HANG THE SYSTEM
UNTIL I PRESSED CONTROL/C.  ONCE I HAD ISOLATED THE STATEMENT AT FAULT,
I HAD EIGHT USERS IN MY POWER :-}

IN CASE YOU ARE INTERESTED, THE INTERPRETER HAD A BUG IN IT'S PARSER.  THE
STATEMENT:  "FORX9=1TO9\NEX9" LEFT THE PARSER IN AN INFINITE LOOP.

SURPRISINGLY, THE TEACHER (ONCE HE FOUND THAT I USED THIS WITH RESTRAINT)
WAS PERSUADED TO LET ME AT THE FRONT PANEL AFTER SCHOOL.  (HIS MISTAKE
AND MINE)

    I used to work on a PDP8/E when I was a research student.  When
    the machine was delivered (a little before my time), it had two
    9 track tape drives as its mass storage.  Someone wrote a file system
    for these - random access on pre-blocked tapes!!!!!  When I met
    it, it had an RK05.  We had an FPP-12 (transistorized) on it, which
    handled all those lovely 36-bit real numbers, and left the 8 itself
    free to do useful things like running the lights round in a loop!
    
    My hacking consisted of writing a "device driver" (we didn't call
    it that, but I suppose it was!) for a "terminal" that consisted
    of a teletype for input, and a Tektronics 611 point-mode storage
    scope for output.  This driver was patched into the Fortran run-time
    system in place of the code for doing floating point on the 8 if
    you had no FPP.
    
    While on the subject, we haven't mentioned the good old cold boot
    of the machine from disk (assuming that a simple start at 7600 wasn't
    enough because memory had gone).  It was
    
    		0030	6743
    		0031	5031
    
    and start at 0030.  For anyone who doesn't understand, opcode 6743
    started a read from disk (from block 0 into location 0, since CLEAR
    had just been pressed), and 5031 is a jump to location 31.  Therefore
    you kicked off a read from disk, then did a jump here, until the
    data from disk clobbered the instruction you are executing!!!  What
    machine today can be started from nothing so simply!  See what we
    lose in the name of progress.
    
    And remember all those lovely instructions for setting the accumulator
    to a particular value - I haven't got a programming card to hand,
    and I can't remember them.  Perhaps someone else can elaborate.
    Things like CLR,INC,RAL in one instruction to set the accumulator
    to 2!
    
    While on the subject, let's remember CCL.  The basic command language
    at the console (whoops, sorry!  Teletype) was very simple.  If it
    came to a command it didn't understand, it chained to CCL.  Clever!
    
    In CCL was a MAKE command for creating a new file with TECO.  There
    was also code to "special case" the command MAKE LOVE so that it
    typed "NOT WAR?" instead of creating the file!!!!
    
    Ah!  Then there was the date (it had to be entered manually and no
    time).  12-bit words gave enough for a date in one word.  That was 5
    bits for the day, 4 for the month and 3 for the year.  The year
    was biased with 1970, and so "ran out" at 31 Dec 1977.
    
    Only that I miss having the SENSE SWITCHES and LIGHTS next to me while
    working!
    
    T

re .38

I remember  COMPIL.MAC  having  the  make  love,  not  war hack in it with a
comment by R. Clements to leave it in because it was a good sales gimmick..

Harley

	    This was also part of TECO, back before it became supported,
	one would use a MAKE command to create a file with TECO.  The
	command has been replaced with EDIT/TECO/CREATE, and it doesn't
	do anything special if the filename is "LOVE."
	    Royal bummer.
			<_Jym_>

    
    	One of the TECOs I used to use could be invoked via the MUNGE
    command to run a macro file. Very descriptive command name, that!
    
    
    Trev

    TECO-11 still supports the MAKE command, and, of course, the file
    "LOVE" is special-cased. The neat thing about this is that the MAKE,
    TECO, and MUNG commands are all parsed by a TECO macro (TECO.TEC),
    and the code to print "Not war?" is implemented in TECO.
    
    Kevin

	    [RE .41]:  MUNG is the appropriate verb.  It means "MUNG
	Until No Good."
			<_Jym_>

    $ MAKE :== $TECO MAKE
    $ MAKE LOVE
    Not war?
    *
    
    VAX/VMS V4.2
    TECO-32 V40(?)
    
    tom_p

	    Hmm.  IMAGELOG reveals that both MAKE (the command symbol)
	and EDIT/TECO/CREATE run the same TECO.  So it's still there!
			<_Jym_>

Re : .31 Yes, I used thousands of DECatrons. A gas filled valve with an 
arrangement of 3 lots of 10 wire electrodes. By appropriate 
differentiating circuits one could create a 3 phase signal which caused 
the glow on one electrode only to step around in 10 steps, giving an 
output pulse each revolution. Often used in nuclear counters, etc.

Re: .25 Should be able to emulate a PDP-8 on the VAX. Seen my CPMAME for 
emulating a 8080 system running CP/M. Its in the toolshed.

My first machines were Scientific Data Systems (later bought out by 
Zerox) SDS910 and SDS920 The 910 had 4k x 21 bit words, and the 920 had 
8k. One of the first to really use vectored interrupts - had about 4k 
interrupts available. Used to run NASA ground stations for many years.
Thought it was a great day when we got our first paper tape reader - no 
punch, you took this little metal block in your left hand and with this 
little punch in your right hand you hand punched each and every hole by 
hand.

My first PDP-11 was an 11/20 ser. no 310. Digital used to keep borrowing 
it back from me all the time to demonstrate it to other customers with 
the promise of another months free maintenance. Had free maintenance for 
about 18 months on it. When lightning struck it one day, there were two 
faults - one in the CPU and one in the memory. Fault in CPU was a 74H74 
flip flop with one half gone. None available in australia so just 
pushed a 7474 over the top of it so the legs contacted and all ok. That 
was a common method of checking for faulty chips then without having to 
remove them from the board. Memory fault was another 74h74 but luckily 
there was a spare half chip on the board about 9 inches away, so half a 
dozen long jumpers and good as new. Ever wondered just why the PDP-11 
had the instruction set that it did. Just look up the data specs for 
74181 ALU chip and suddenly all is plain. The old 014747 instruction was 
good for more than a laugh - it was one of the quickest ways to check 
out all memory locations in no time flat. Because the registers appeared 
as locations on the unibus as well, you could use memory referencing 
instruction to get at the registers, or even better, put some 
instructions into the registers and execute them at VERY high speed.


	Enough nostalgia -

		Doug.


(REAL programmers program on the switch panel..................)

    	Congratulations to .46 on the DECatron. I was thinking that
    no-one around was older than me :-)   I saw them in action, but
    never had a chance to design a circuit round them.
    
    		Dave

    You can also use them as up-down counters. I still have some at
    home. Initially intended for the 'accumulator/display' for a rallye
    computer. At that time (early 60's) it was the cheapest way to get
    a bi-directional counter and display that would run at 1 Kc.
    
    	Wayne
    

    jumping in late on conversation....
    
    it is interesting to note that 8's did most of the hardware design data
    postprocessing for the processors you are probably running on or
    at least through...
    
    The CAD department in the mill was a big user of 8's through at
    least 1980. They were used to post process CAD data into: plotter
    tapes to plot the Star and Comet PC board etch, drill tapes to drill
    the PC board holes, and automatic component insertion tapes to stick
    the components into the holes!
    
    Remember the "datacases" ? briefcase-like things with blu plastic
    molded compartments for dectape & papertape?
    
    
    Yeeeechhh, never again will i shoehorn a program into such a small
    machine!
                            
    
    paulsears

    We has a PDP-8/E[?] in high school running TSS-8.  STarted out with
    2 terminals [Teletypes, with paper tape punches, of course!], one
    of the first things I figured out how to do was punch character
    shapes into the paper tape.  One of the terminals was in this little
    concrete room about 4X the size of a phone booth, I have memories
    of finally turning the TTY off and hearing pounding in my ears for
    long periods of time thereafter.  Somewhat later we got [WOW!] three
    KSR Decwriter IIs [high speed!].  I have a picture somewhere in
    my files of myself at work at one of them with a 3 inch stack of
    paper piled up in the wire paper-catcher....  A further upgrade
    was an ADDS video terminal that ran at (I think) 2400 baud, boy
    was that ever in demand.  The next upgrade they did was to get a
    bunch of Commodore PETs and network them together, which effectively
    obsoleted the PDP-8.  For me that was a great big step backwards,
    but I guess they wanted more than 4 kids at a time to have access
    to a computer...
    	One of my favorite programs was a rather lengthy BASIC program
    that combined the functions of a rudimentary text editor, a string
    scrambler, and an optional byte substitutor.  Used to use it for
    a private EMAIL system/BBS.  Back in high school, with the girls
    school 5 miles away, this was important!  :+)
    	At a job I had quite a few years back, my first exposure to
    the innards of a computer was when we had to fix a PDP-8/I that
    was supposed to run a component tester.  We found:
    	1)	Of 4 fans in the bottom of the rack, 2 were completely
    stopped, one was turning at half speed, and the other was about
    3/4 speed.
    	2)	The CVT resonant capacitor [condenser? :+)] had drifted
    in value, and the power supply had a glitch every half-cycle that
    caused the 5 volt supply to drop to 2 volts.
    	3)	Of course none of the front panel lights worked anymore...
    Apparently this machine had been left running for a couple of years
    without even any P.M.....
    	4)	Once we got that all straightened out, we could begin
    to debug the CPU.  We had 4 different print sets, lotsa lotsa little
    cards with things like '6 D-type Flip-Flops' on them, and wires
    all over the place....  I distinctly remember looking at the TTL
    in the system and wondering why none of the clock lines or other
    signals even made it up to 2 volts, wasn't 2.4 the minimum?  I was
    told that they run faster if you don't pull them too high, is this
    correct?  Turned out to be (I think) one of the aforementioned D
    type flip-flops, only took us about 3 days total to find the problem,
    including fan/power supply/front panel...  Not bad considering neither
    of us had worked on a computer before.
    
    WPNS
    

    Re: .50
    
    I've seen my share of those.
    
    I think the most difficult machine I worked on was the PDP-1. Couldn't
    find parts for it, but it ran a good FOCAL compiler.
    
    Jeff
    

    I knew that I would find this topic in notes somewhere!
    
    My introduction to computing was "The Small Computer Handbook" thrust
    into my hand at a exhibition.  About a year later an application
    came up that would benefit from the inclusion of this remarkably
    cheap computer (sorry, that should read processor.  We were not
    allowed to use the word computer in any paperwork in case the clients'
    computer services people tried to horn in on the project.).
    
    So we designed the system using a basic, 4K, PDP-8L, complete with
    ASR33.  We found some third party terminals (DEC didn't offer any)
    and a fixed head magnetic drum and bought a lot of logic boards
    from DEC to make a system to control a real time system, 24 hours
    a day, 364 days a year.  We were allowed one hour a year for
    maintenance!
    
    Development software consisted of an editor and an assembler on
    paper tape.  We got a skeleton interupt handler from DECUS.  We
    had never heard words like "operating system" or "high level language".
    We did get fed up with toggling in RIM every time we had a crash
    so we put the editor and assembler onto the drum and devised a short
    bootstrap to load the top page of memory with what would now be
    called a rudimentary operating system.  The paper tape sources still
    sit in a drawer at home.
    
    The application, after much shoehorning, came to around 12K of code.
    Naturally we had to partition it and run with much overlaying.
    
    You hackers might be interested in a subtle and ellusive hardware
    bug that the machine was shipped with.  It took 3 months to pin
    it down to a tight wait loop in high memory flipping a bit on page
    zero.  We changed the core stack; we changed every board in the
    system, we changed the power supply; we changed the card cage and
    chassis.  The fault stayed!  Why??
    

    When the next application came along, we knew a bit more and technology
    had advanced somewhat.  We were determined not to be caught short
    on memory again so we splashed out on 8K of core and a massive 256K
    on the drum (those drums were very reliable).  Of course this
    application was much bigger!
    
    Would you believe - 4 interactive users on top of multistation bit
    pattern generation and testing at a clock rate of 1khz.  The
    application program included an operating system, a development
    environment and a language all designed from scratch.  Did I mention
    the file system.
    
    Development was a bit easier.  We used OS/8.  We also used a bootstrap
    ROM.  I programmed it myself...with a pair of wire cutters.
    
    The system took 3 years to design, develop, and commission for a
    total planned usage of nine months.  The pay back time was just
    6 weeks.
    
    This one had a hardware bug too, but then what is a 5 nanosecond
    race hazard between friends?
    
    A thought has just occurred to me.  Memory is like your income -
    no matter how much you have, it is never quite enough! (see VMSTUNING)
    

    re: .52
    
    	The old PDP8s (not sure about the 'L') if you addressed one
    particular address too often the magnetic cores would warm up to
    the point where they dropped bits. I think it took about 5 hours
    for an infinite loop (JMP .) to complete. A longer infinite loop
    would take longer, of course, and a really long infinite loop would
    take forever  :-)

    Re: .54
    
    I think that "warming up the cores" may have had a bearing but the
    loop was an 80 second timeout.  The cause was deterministic and
    fixable.

    	No, ours definitely took more than 80 seconds. No-one told us
    it was fixable, but then we were not too bothered since we normally
    wrote longer loops than that. Maybe we had a colder computer room?

    
    			PDP-8....  I repaired "L" and "E"...
    
    	The "Hot Memory Diag" did and does exaclty that.
    
    	If you hit a CORE STACK area repeatitly you will heat-up
    	the memory array...  If you allow the test to run TOO-LONG
    	you will destory the STACK, as per the warning on the Paper-Tape.
    
    	Same test for PDP-9, and PDP-15...
    
    	If you are getting failing bits....
    
    		Try this....   Remove the CORE STACK and
    		paddle boards.
    
    		Get a bucket (large enough to put the CORE STACK in...
    
    		Now fill the bucket with WARM/CLEAR water...
    
    		Dip the CORE-STACK into the water (many times),
    		and watch for dirt to wash-out...
    
    		Afterwards....
    
    		Allow the CORE to dry (SLOWLY).... Don't use HEAT-GUNS!
    
    		(I use to work in Las Vegas, only 30 minutes in the
    		 SUN).
    
    	I realize this sounds VERY WRONG, however it never hurt any
    of the units and always repaired the CORES....
    
    
    	If you wish...  Call me (553-3240)
    
    	Al