My pdp8/e reference card, circa 1970
[attachments too big for wysiwyg apparently...]
Last Edited: Sun. Mar 5, 2017 - 08:20 PM
How many "powerful" instructions?
Author
Message
second side
Attachment(s):
You're showing your age! I've currently got its bigger brother, the lsi11/23 on my desk at the moment. I can't begin to tell you how much of a challenge it is designing an interface to its open collector bus when the ics are obsolete!
An interesting factoid - the pdp8 had no means of loading a 12bit constant, so a 'literal pool' was used. Fast forward to today and the cortex m0 uses the same technique.
Level: Moderator
Joined: Mon. Jul 18, 2005
Posts: 101882 View posts
Location: (using avr-gcc in) Finchingfield, Essex, England
Kartman wrote:
I am sorry that I did not salvage those chips when I discarded my 11/23+. I did save the SASI/SCSI host adapter I designed for it which might even still work. Here are the top and bottom:
... I've currently got [an] lsi11/23.... I can't begin to tell you how much of a challenge it is designing an interface to its open collector bus when the ics are obsolete!...
It had an interesting bootstrap mechanism that made good use of the PDP-11 instruction set.
That's how we did it 30 years ago. I can get the required bus interface ics - new old stock at around $6 a piece. I want to make something from currently available parts, so i've settled on am26s10 devices. I'll probably throw in some lvds receivers to get the correct rx logic levels. Along with a Lattice machx02 cpld/fpga should give me a nice bus interface to hook a beaglebone black up to. I hope to be able to emulate both a cpu and peripherals.
The above board is from 1984. I don't feel any nostalgia about manual wire-wrapping. 
Hello all!
In order to preserve some of my mental health, I have given up answering the same ol' same ol' here. I risk it to make this impromptu appearance since Lee's images above made me feel a bit young (er). The VAXen was what I grew up on, so PDP-8 was before my time. I just love how there are CPU instructions for doing punched paper tape IO!
Posting this also reminded me how badly the site UI is on a pad. So...
Now, back to sanity. Take care, everyone!
Last Edited: Mon. Mar 6, 2017 - 03:34 PM
JohanEkdahl wrote:
The VAXen was what I grew up on, so PDP-8 was before my time.
You whippersnappers.
jfiresto wrote:
I don't feel any nostalgia about manual wire-wrapping.
As an undergrad, I hung around what would now probably be called the "micro lab" with an assortment of undergrads, grad students, professors, and support staff. Some of my projects were pdp8 interface boards -- one was essentially a bus extension with single-step simulator/emulator capabilities so yet others could test their add-on extension boards. [that era's equivalent of Arduino shield?]
Luckily I had a very dexterous girlfriend who enjoyed soldering and wire-wrapping, and was quite good at it. Really helped me out.
A significant part of lab budget was spent one year on a 4K (probably 12-bit words?) core memory add-on board for pdp8. About $16000 IIRC -- $4/word.
After college I went to work for Burroughs on mainframes. In 1974 they started selling semiconductor memory add-ons for $10000 per megabyte. I was sure they would go broke at those prices.
Level: Moderator
Joined: Mon. Jul 18, 2005
Posts: 101882 View posts
Location: (using avr-gcc in) Finchingfield, Essex, England
theusch wrote:
A luxury we didn't get in the UK :-(
a very dexterous girlfriend who enjoyed soldering and wire-wrapping
The first computer that I used on a professional basis was a RAD-8, which had a pdp8 at its core.
Interesting to note that the pdp8 had IOT. Everything old is new again!
The PDP-11 was the first computer I used/programmed on a professional basis. I believe it was the 11/73.
Interesting to note that the pdp8 had IOT.
It's got nothing on the CDP1802! The 1802 had SEX...
I've been cleaning the garage, and came across this just a couple days ago...
PDP11:
https://www.flickr.com/photos/58...@N00/33293948355/in/album-72157681026854386/
Level: Moderator
Joined: Wed. Mar 28, 2001
Posts: 30169 View posts
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweis, but plenty of Corona beer)
The 1802 had SEX...
With the 6809?
What have I started??
Level: Moderator
Joined: Wed. Mar 28, 2001
Posts: 30169 View posts
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweis, but plenty of Corona beer)
Well the 6809 has the Sign EXtend B accumulator into A accumulator instruction....opcode 0x1D
I'm intrigued by the PDP8's 'OPR' instruction. Seems a very powerful, if somewhat hard to program, feature.
I recently read a biography of Alan Turing, https://global.oup.com/academic/... , which prompted me to buy another book which goes into more technical details on the computers he designed and which followed, https://global.oup.com/academic/....
What stands out is how different the architectures were from what we have today. A lot of this is down to them being bit-serial machines and having to use Mercury delay lines as temporary stores (each of which held multiple words a bit like a long shift register) but the genius behind their design is quite amazing. Consider Turing's Pilot ACE and its instruction format; each instruction contained the following...
*) Source of next instruction
*) Source
*) Destination
*) Count
*) Position of next instruction in shift register
The sharp eyed will have spotted that there IS NO OPCODE in there.
Every instruction worked the same; it transferred 'count' words from 'source' to 'destination'. That's it. The clever bit is that the operation to be carried out is implicit in the source and destination used. To use a couple of examples from the above book, if the source is S21 then the data transferred is the logical sum of temporary stores TS26 and TS27...
S21 = TS26 & TS27
Or, if the destination is D17 the the data written to that location is the result of adding TS16 to the source...
D17 = S + TS16
So the instruction...
DL7/S21/D17/1/2
...adds the contents of TS16 to the logical sum of TS26 and TS27 and transfers the result to D17. The next instruction comes from delay line 7 and is the second word in the bit stream.
Quite amazing.
I can't help thinking that semi manufacturers are missing a trick with the processor architectures.
To anyone even a little bit interested in processor architecture I highly recommend you to buy the second book above from your local independent bookstore. And for a good old-fashioned read, the first one.
Last Edited: Tue. Mar 7, 2017 - 08:33 AM
Kartman wrote:
How about this list of 'powerful' instructions:...
It appears to be missing one...
ROJ - Randomise Opcode and Jump
Brian Fairchild wrote:
Kartman wrote:How about this list of 'powerful' instructions:...
It appears to be missing one...
ROJ - Randomise Opcode and Jump
And it doesn't have HCF (Halt and Catch Fire) either. ;-)
S.
Scroungre wrote:
Actually, it is there...
And it doesn't have HCF (Halt and Catch Fire) either. ;-)
frog_jr wrote:
Scroungre wrote:Actually, it is there...And it doesn't have HCF (Halt and Catch Fire) either. ;-)
Yes, it does. I looked and missed it. You are correct, it's in there (among a few others). S.
After i'd figured out how to write sc/mp machine code (signed math had me stumped for a while), i then wondered how to build a processor. Unfortunately i had no access to reading material that would teach me since the local library had no books on the subject. Since then I've been interested in computer architecture. Recently I read a book on the CDC6600 supercomputer by Thornton. 400,000 transistors and 150kW! Not something you'd make in your backyard.
Mercury delay lines were mentioned - there were wire delay lines and magnetic drum storage as well. The 50's and 60's seem to be the golden age of engineering.
Kartman wrote:
...I've been interested in computer architecture...
Ever since I went and saw the Megaprocessor last year I've had an urge to build something myself. Not on the scale of Jame's creation but something smaller. And I keep being drawn to the idea of a bit-serial machine or something very off-the-wall.
I'm lucky in that one of my jobs often has me kicking my heels for a couple of hours in London, just around the corner from the Science Museum, where this resides...
...that's the Pilot ACE, a bit-serial machine using valves. Now, each valve can be replaced by 1 or 2 FETs so replicating this...
..wouldn't end up that big. There are emulators out there so there's likely to be enough info to build one.
I spent the weekend re-arranging my 'metal-bashing' workshop so my CNC machines now have a permanent home nearer my office...
...so it's all coming together.
Level: 10k+ Postman
Joined: Tue. Jan 9, 2007
Posts: 13769 View posts
Location: Clay County, Texas, u.S.A.
Brian Fairchild wrote:
Turing machines are a part of a formal methods course.
I recently read a biography of Alan Turing, ...A lot of this is down to them being bit-serial machines and ... but the genius behind their design is quite amazing.
Brian Fairchild wrote:
May not be of interest to the ones with fabs whereas would be of interest to some that are fab-less (i.e. niche)
I can't help thinking that semi manufacturers are missing a trick with the processor architectures.
There are power advantages for an asynchronous architecture (however fast the serial bits arrive but not too fast)
Your design can be modeled on your PC then fab'd by MOSIS.
My initials are an instruction JSP Jump Set Parody
Back when I was in school, the 2 year college could not afford a PDP-11 or a Vax. We had an HP2000F Time share BASIC system. This could handle 32 users at once. OS system commands were 3 letters. only the first 3 letters were parsed the rest were ignored. I think you logged in with the HEL(LO) command. There were also the usual basic commands such as RUN, LIS, and PUN. The latter enabling the tape punch in the teletype.
My favorite command was the POR(T) command, Which also was the firs letters of my last name. This was an admin command, that would list the active user and when entered from a user account would generate the error "Privileged command."
I used to get other students by having them enter their name last first which would result in the ??? error. When I did it the computer claim that I had "Privileged command."
In the electronics department (I took an additional 2 year trade EE certificate.) Was given a PDP-8 console. It smelled bad. When we opened the access panel there was a dead rodent inside. Someone remarked that "Well there's the problem, the Squirrel died." This got me thinking that if Grace Hopper had found a squirrel instead of a moth, and that the plotting planchets were called bugs, that computer errors might have been called Squirrels, and Mice called bugs.
I still have a CARDIAC, with it's 10 most powerful instructions and the cardboard bugs. Those 10 instructions would probably be the ones Ada and Babbage used. They are 3 digit Decimal op codes. Actually those of us who have seen the drawings and looked at Ada's paper, They never got past the microcode part. On the other hand you could code anything into those rotating barrels so some time I want to build a 3 digit engine using the Cardiac instructions.
As for wire wrap. I am wire wrapping some Ardino shields today. Will post some pictures ...
jporter wrote:
I still have a CARDIAC, with it's 10 most powerful instructions and the cardboard bugs.
I figured you would chime in here... ;)
I've been playing around with the PiDP-8 kit here:
http://obsolescence.wixsite.com/...
It has been an interesting learn. It uses a raspberry pi running the SIMH simulator and a custom panel to show all the blinky lights.
It is a quite odd little machine with a word size of 12 bits. I recently ported some of my Enigma code to basic to run on it. It has no stack and isn't well suited for C though sadly.