Some More Timesharing Documentation, Plus History

[Unibus]

Hi,

The Hospital Computer Project, Time-Sharing Executive System, BBN Report Number 1673, April 1968

TABLE OF CONTENTS                                           Page
List of Figures ...........................................   iv
List of Tables ............................................    v
Preface ...................................................   vi

     Introduction .........................................    1

  I. The Hardware Environment .............................    5
 II. The Swapper ..........................................   15
III. The Dispatcher .......................................   30
 IV. The I-O Processor ....................................   35
  V. Teletype Service Routine and Other "Slow" I-O ........   46

LIST OF FIGURES                                             page
Figure 1. Interconnection of processors
          and memory banks ................................    8
       2. Round-robin queue ...............................   16
       3. Multilevel queue ................................   17
       4. Swapper flow diagram ............................   26
       5. Item format in core memory ......................   38
       6. Block and item formats on Fastrand drum .........   38

LIST OF TABLES                                              page
       I. Interrupt priorities ............................   11
      II. Buffer allocation for slow I-O devices ..........   47

Linked to the interrupt system are the hardware components which protect the computer system from user-issued "privileged" instructions or memory-bound violations.

                         TABLE I. Interrupt priorities.

Priority  Source of                                 Report Section

(Octal)   Interrupt                   Describing Routine

0         unused (highest priority)

1         high-speed data channel     I-O Processor

2         paper-tape reader           Teletype, etc.

3         line printer                Teletype, etc.

4         I-O controller commands     I-O Processor

5         program swapping drum       Swapper

6         term  inal scanner          Teletype, etc.

7         one-second clock            Dispatcher

10        one-minute clock            Dispatcher

11        unused

12        paper-tape punch            Teletype, etc.

13        unused

14        console typewriter          Teletype, etc.

15        I-O processor program       I-O Processor

16        privileged-instruction      Dispatcher

          trap

17        32-millisecond clock        Swapper

So it is no wonder there is confusion over PDP-1D Interrupts and its hardware configuration.

From Conversations with Medical Informatics Pioneers: 2004 Interview of Octo Barnett

OB ... We couldn’t use telephone lines, because at that time it was all plugin, operator-type stuff. And so we had to get a solid copper wire connection.
DS Between Cambridge and—?
OB Between Cambridge and here [MGH in Boston]. Well, the telephone company didn’t have any of those. They didn’t have a thing called a tariff. They had no way to sell it. And this engineer [Jordan] found out, basically, that there was a tariff for lines that were laid along a railroad track, using the telegraph wires, because for the telegraph wires, you had to have a continuous connection.

[Jordan] said, “Why don’t you pretend you’re going to lay up railroad track between Cambridge and MGH?”

DS Good idea.
OB So they put a copper wire underneath the Charles, and got over here, and we got four copper wires connected over there.
DS Just for you?
OB Just for us, yes.

DS JA [sic] Who paid for that?
OB I don’t know.
DS JA [sic] He just got it done?
OB With Jordan, we would never know what he paid for what.

From A Culture Of Innovation, Insider Accounts Of Computing And Life At BBN

According to Fredkin, when McCarthy explained his ideas for time-sharing, he said that time-sharing should be done in RAM with interrupts as is done today. However, at the time no one could afford a big enough RAM: it was $1 per bit then, which would be $6 or $7 per bit now. Fredkin had an idea about how to demonstrate that McCarthy’s idea was right: Fredkin invented the swapping drum. The basic drum came from Vermont Research. The PDP-1 initially had 4,096 18-bit words of memory. In Fredkin’s design, the swapping drum could read in 4,096 words while simultaneously writing out another set of 4,096 words in 20 milliseconds (5 microsecond cycles). He studied the timing diagrams of the computer RAM read-write cycle and the drum read-and-write timing and figured out how each read-write cycle of the computer could read one word from memory to the drum and then write one word from the drum to memory. Furthermore, because the drum was 4,096 words around and memory had 4,096 words, it was possible for the system to notice where in the 4,096-word interval the drum was relative to its read-write positions and to start the transfer at that same point in the 4,096 words of memory; in other words, there was no latency waiting for the drum to spin to the beginning of a 4,096-word drum block. However, there was a problem. To be synchronized with the machine and have 4,096 words around it, the drum need to rotate at 3,000 rpm. Unfortunately, Vermont Research couldn’t find a motor of the right design and right speed, so the actual system ran a little slower than Fredkin’s optimized design. In Fredkin’s view, it is unfortunate that, rather than seeing his use of such an optimized swapping drum as an indication of what a time-sharing system could do if it was all in RAM, the whole world copied the idea of a swapping drum but without having a drum that worked like Fredkin’s. Thus, later time-sharing systems had lots of latency and were very slow.

Fredkin continues,

The hardware suggestions were mostly in the PDP-1 before it arrived. However, the swapping drum was added later. It took quite an effort to convince Digital to do it. There is a great story about that event. The second PDP-1 went to MIT, where Professor Jack Dennis led a group of students who implemented a lot of good software. He also wanted a “swapping drum” to do time-sharing. I kept pestering Gurley to offer to build it, but he never got back with a proposal. One day, McCarthy, Gurley and I were all at MIT and John and I suddenly started pestering Gurley to agree to build the drum system I designed. Gurley’s response was that we hadn’t ordered it. John and I both said something like “You mean, if we order the swapping drums right now, then you’ll build them?” Gurley laughed and said “Yes.” John said “Wait right here.” He ran down the hall to Professor Zimmerman’s office (I think) and got them to give him a PO number from RLE at MIT while I got on the phone to Licklider and asked him to get me a BBN purchase order number. Amazingly, in about 15 minutes, Gurley had two PO numbers and agreed to build a swapping drum system for both MIT and BBN, which he did.

...

Also according to Cole, Model 28 TTYs with 5-bit Baudot code terminals later replaced the Sorobans, and still later Model 33 TTYs with 8-bit ASCII were used.

Mann continues,

The most exciting story was the time, near the end of the project, when someone decided that we needed a computer center manager and appointed Louis (Lew) Clapp, who was a noncomputer scientist from an acoustics group. DEC had had a lot of trouble with that PDP-1 (it was really a prototype) and frequently engineers or techs made wiring changes, which they carefully noted on a huge set of prints [of the PDP-1 wiring]. One Friday evening they went home, leaving the prints spread out on the floor at the back of the computer room; Lew came in, saw them, and threw them out, on the theory that people who made a mess in his computer room should be punished. He was immediately fired. A few weeks later he was rehired by the acoustics people.

...
Exec II (BBN’s second time-sharing system) was written for the 1d, led by Nancy Haggerty, according to Jon Cole. However, Exec II never worked reliably. Thus (still according to Cole), Steve Weiss, Andy Munster, and others, planned to build a new, much more cleanly organized time-sharing system, known as Exec III, with state table for users, etc. The Exec III time-sharing system was extensively documented, and the PDP-1d and Exec III ran for a long time after the end of the hospital application project, residing in the back room of BBN’s 20 Moulton Street building.

The reference document to EXEC III is  The Hospital Computer Project, Time-Sharing Executive System, BBN Report Number 1673, April 1968, detailed above.

Regards,

Garry

[Bill E]

That's a nice find. Yes, one thing about trying to resurrect various devices is that it's become clear that there really wasn't a lot of standardization. Random opcodes, devices, interrupts, etc. were common.

I guess that points out just how flexible a platform the -1 was. The -1D's seemed to be the most hacked, especially #45, the BBN one that implemented some, hmm, interesting bits.

It has devices and instructions unique to just it. The PDP-1X, #38, at MIT went even more bonkers. It added 24 or so new instructions and a completely separate interrupt system, 'traps'.

It seems BBN adopted/adapted many of the ideas for their #45 mods.

Bill

[Whit Turner]

These are great stories! Imagine running copper cables under the Charles....

Whit

[Bill E]

Well, MIT also rotated a poor guy head over feet all the way across the Hahvahd bridge, hence the distance measurement  'smoot'. Copper was probably a bit easier, just walk over the bridge unrolling a spool into the water. Of course now a swat team would show up and shoot you.

Bill

[Glenn Babecki]

In fact 364.4 smoots long, plus or minus one ear.

On Mon, Nov 3, 2025, 9:00 AM Bill E <wjegr...@gmail.com> wrote:

Well, MIT also rotated a poor guy head over feet all the way across the Hahvahd bridge, hence the distance measurement  'smoot'. Copper was probably a bit easier, just walk over the bridge unrolling a spool into the water. Of course now a swat team would show up and shoot you.

Bill

--
You received this message because you are subscribed to the Google Groups "[PiDP-1]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-1+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pidp-1/17e3ccd7-19b5-43d7-8d83-9b91c3da6d58n%40googlegroups.com.