Listing all known ITS machines

[larsbrinkhoff]

This is what I know so far.

At MIT.

Model	Serial number	Name	Comment	Owners
PDP-6	2		AI Lab	MIT (1964-1982)
PDP-6			Dynamic Modeling	MIT (1969-)
KA10	8	AI		MIT (1968-1983), Concourse (1983-)
KA10	144	DM	Dynamic Modeling	MIT (1970-1983)
KA10	198	ML	Mathlab	MIT (1971-1984)
KL10	1038	MC/MX	Macsyma Consortium	MIT (1975-1988), Peter Löthberg, LCM
KS10	4627	AI		MIT (1985-1990), Christoper Zach, LCM
KS10	4648	MD	Mostly Development	MIT (1986-)
KS10	4649	MX/MC	Mail Computer	MIT (1986-1990), Christoper Zach
KS10	4653	ML		MIT (1986-1990), Christoper Zach, Doug Humphrey, Dave McGuire

Outside MIT.

Model	Serial number	Name	Comment	Owners
KS10	4428			Peter Löthberg
KS10	4469			Gordon Greene, Daniel Seagraves
KS10	4655	LI	eLinor	LIU, Lysator, Ludd, Anders Magnusson
KS10	4664	PM		Marc Crispin, LCM
KS10	4142	ZP		John Wilson
KS10		SI		Stacken
KS10		FU		Flinders University
KS10		DX	Digex	Doug Humphrey (1989-)
PDP-10/X		TX		Dave Conroy
KLH10		SV		Paul Svensson (2001-)
KLH10		UP	Update	Björn Victor (2004-)
KLH10		ES		Eric Swenson's ITS (2015-)
KLH10		NO	nocrew	Lars Brinkhoff (2017-)
KLH10		SJ	San Jose	@b4 (2017-)
KLH10		XM		@mattwyrm

---

Other PDP-6'es in Cambridge:

Model	Serial number	Location
PDP-6	6	Keydata (timesharing company at Tech Sq)
PDP-6	8	MIT Laboratory for Nuclear Science

---

TOPS-20 machines at MIT:

Model	Serial number	Name	Comment
KL10		XX	1978- DM users moved here
KL10		OZ	1982-1988
KL10	2460	EECS
KS10	4380	BLT, LSD	Brave Little Toaster
		MIT-SPEECH

WAITS at Stanford:

Model	Serial number	Name	Owner
PDP-6	16	SAIL	Stanford 1966-1980
KA10	32	SAIL	Stanford 1968-
KL10	1075	SAIL	Stanford -1991
F2/F4		CCRMA	Stanford -1990
F2		S1-A

[larsbrinkhoff]

Sources:
SYSTEM; CONFIG >
SYSTEM; SALV >
https://sites.google.com/site/mthompsonorg/Home/pdp-10/pdp-10-serial-numbers by @mgthompson
http://www.ultimate.com/phil/pdp10/pdp6-serials.html by @budd
ftp://ftp.ultimate.com/pdp10/bucs20-anon/decapr/decapr.cpu.txt
Project MAC Progress Reports from dtic.mil
Messages to KS-ITS
Messages to ITS-BUGS
Messages to ITS-LOVERS
http://infolab.stanford.edu/pub/voy/museum/pictures/AIlab/SailFarewell.html

[larsbrinkhoff]

Questions:

• Did the KS10 ML become DX?
• Did SI come from a previous MIT machine?

[larsbrinkhoff]

Stacken apparently tried to get ITS running on KATIA in 1985. But the SI machine known to CONFIG is a KS10.

Per Lindberg was at Stacken at the time; I've asked him for details.

MRC wrote in 2001:

I also salvaged the [SAIL] BBN pager; the last I heard, Peter Lothberg & co. had it on a working KA10 in Sweden running Tenex.

[mgthompson]

Getting ITS running on my KS10 is on my to-do list. I made TOPS10 boot tapes using the HP SCSI 1/2" tape drive on my SPARC server, and the worked fine on the KS10's TU45 tape drive. The power supply in the HP tape drive failed, I repaired it, and now the KS10 cannot read tapes made on the HP tape drive. The HP tape drive can easily read its own tapes, and tapes from the KS10.

let me know if you see any missing information in the PDP-10 serial number list.

On Thu, Dec 8, 2016 at 7:07 AM, Lars Brinkhoff notifications@github.com wrote:

Stacken apparently tried to get ITS running on KATIA in 1985.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/PDP-10/its/issues/181#issuecomment-265725315, or mute the thread https://github.com/notifications/unsubscribe-auth/AEFUHo1_dr145LEnxCnq4MIlKGwWtJjpks5rF_LzgaJpZM4LHmwI .

-- Michael Thompson

[larsbrinkhoff]

@eswenson1 Did you edit the table? I didn't notice, so your changes are not present in #259.

[eswenson1]

I had thought I'd added my machine to the list. But noticed it wasn't there when you committed -- but also noticed your comment about emulated systems and figured you didn't want them there. (Even though I've had ES-ITS up for over a year! :-)). No matter.

[larsbrinkhoff]

Fixed by #259.

[larsbrinkhoff]

There's a rumor there was a DynaMod PDP-6. (And I have some documents to back it up.) @eswenson1 or anyone else, have you heard of such a thing?

[eswenson1]

Well, I don't have any idea. But DynaMod sounds like "Dynamic Modeling", which is what DM stood for. So given that we had all sorts of PDP-this-and-thats attached to various of the MIT ITS machines, I wouldn't be surprised, if there was a PDP-6 attached to DM....

[larsbrinkhoff]

Ok, thanks.

Right, DM = DynaMod = Dynamic Modeling. The PDP-6 was the predecessor to the PDP-10. The AI Lab got started with ITS on a PDP-6 in the late 60s, but replaced it with a PDP-10 which was a much more reliable machine.

PDP-this-and-thats would have been PDP-8 and PDP-11 minis. The 6 was a mainframe-class machine, so it wouldn't have been attached as a mere peripheral device. Maybe as a secondary CPU running alongside a 10.

[eswenson1]

Ah, right. Was confusing PDP-8s with PDP-6s. I never actually (knowingly) used the 6s when I was at LCS.

[larsbrinkhoff]

I'm reopening this, to add more research notes.

[larsbrinkhoff]

I found some references to another PDP-6 machine used by the Dynamic Modeling group. It's unclear whether it ran ITS.

• Marc Crispin posted to alt.sys.pdp10:

  I did meet the MIT AI and DM PDP-6s

• Project MAC Progress Report VII, page 59:

  at Project MAC, both the Multics GE 645 system and the Dynamic Modeling and Computer Graphics Groups' PDP-6/10 system are network hosts.

• Project MAC Progress Report IX, page 60:

  Work on the Dynamic Modeling System began, effectively, in October, 1969, when a used PDP-6 computer with 32K words of memory was delivered to Project MAC.

[larsbrinkhoff]

@mgthompson Your list has both this

?? ITS MIT-AI Cambridge, MA KA10
?? ITS MIT-DMS Cambridge, MA KA10
?? ITS MIT-ML Cambridge, MA KA10

and this

144 TOPS-10 MIT-DMS, MA KA10
198 TOPS-10 MIT-ML, MA KA10 (Probably the right S/N)

[larsbrinkhoff]

ftp://ftp.ultimate.com/pdp10/bucs20-anon/decapr/decapr.cpu.txt lists Peter Löthberg's KS 4428 running ITS.

[larsbrinkhoff]

DM PDP-6 confirmed by Noel Chiappa:

I don't know about the software run on the two PDP-6's - by the time I arrived at MIT, they were both powered off and never, as far as I know, ever ran again. I would assume that it ran ITS.

I don't recall if the physical remains stayed until the KA's were de-commissioned, or of they were removed prior to that - I suspect they stayed, since they were mixed in with the KA's - in the case of AI at least, wired in together - but don't remember exactly. I don't recall if the I/O bus was shared between the two CPUs on the DM machines, the way it was on the AI KA and PDP-6.

The DM PDP-6 was part of the DM KA 'assembly' - DM was in two rows (front and back) to the right of the right-hand door from the lobby into the machine room. IIRC, the PDP-6 was in the front row, to the right? of the KA CPU. (The back row contained memory boxes - a mix of different DEC memories. I don't recall where the tape and disk controllers were - or the disk drives. I seem to vaguely recall a few boxes to the left of the KA CPU? Maybe there are some pictures of the MAC machine room that will show it.)

[larsbrinkhoff]

ML KA10

Project MAC Progress Report IX, page 86:

The "Mathlab" PDP-10 arrived during the year and became operational in February 1972.

The basic PDP-10 processor and one half its memory (i.e. 128K) arrived in October 1971. Richard Greenblatt and Thomas Knight made an heroic effort to bring up an ITS time-sharing system compatible with that on the Artificial Intelligence Group's machine. The new file system they created was given to the Dynamic Modeling group and tripled their effective disk utilization.

The remainder of the memory arrived in December. In January Systems Concepts installed a pager compatible with the one on the Artificial Intelligence machine. As a result, the Mathlab machine is able to use the latest versions of the Artificial Intelligence time-sharing systems.

AI memo 238, April 1972:

The ITS system is to be used by the Project MAC Mathlab group on their own PDP-10 Computer. This should increase the incentive for real modularity which has been lacking with a one installation system (Actually the Project MAC Dynamic Modelling Group uses a non-paged early offshoot of ITS on their PDP-10.)

Much of the work in setting up the initial Mathlab system is being done by Richard L. Greenblatt.

RFC 366 from 28 June 1972 says "During this period the MIT Math Lab PDP-10 (Network address 198) became a server."

Project MAC Progress Report X, 1972-1973:

The Mathlab Machine went on the network in May, 1972.

Email from GSB, March 9th 1984:

Date: 8 Mar 1984 20:45 PST (Thu) From: Ian Macky

. . . The "internal error" was apparently due to cca's being down. And ML has officially gone west.

But it hasn't reached the horizon yet, so don't dyke it out of the host tables. Its disk has been fixed, and its processor/pager problem is going to be worked on this weekend. While i don't expect it to accept incoming net connections ever, it will certainly need to reach out and write some files.

The last backed up files from ML seem to be from 1983-09-15.

[larsbrinkhoff]

@taa01776 As you can se in previous comments, I found some references to the Dynamic Modeling PDP-6. Do you know what that machine was used for? Was it running some version of ITS? Did it ever?

[larsbrinkhoff]

AI PDP-6

Lawrence Krakauer:

ITS, written by Greenblatt and others, initially ran on the PDP-6, which was delivered to the AI lab on October 6, 1964.

PDP-6 Memo No. 2, MAC memo 191: https://archive.org/details/teco-6

October 29, 1964
A character-based paper tape editor for the PDP-6 has appeared.

Project MAC Progress Report IV, 1966-1967:

THE PDP-6 TIME-SHARING SYSTEM

A time-sharing system for the PDP-6 went into operation in July 1967, and currently provides service to the eight consoles which are connected to the PDP-6.

Our PDP-6 is equipped with a Fabritek memory of 256K directly addressable words with a 2.75 microsecond cycle time. The memory is utilized to avoid swapping user programs and allows large dynamic I/O buffering. The overhead is therefore small enough to allow 1/30 second quantum per procedure. Such a small quantum essentially guarantees immediate response for interaction-limited programs. (In CTSS, response time may be longer than 60 seconds.)

SYSDOC; ITS HISTRY

[1115] 5/27/78 The following items deleted from the AI system (softwarily): 340, DSD, OMX, IMX, N D/A, PDP6, H CLK

Lawrence Krakauer:

The PDP-6 was retained until February 16, 1982, when it became impossible to maintain, and was taken apart. I know these dates because I have a board from the PDP-6 hanging on my office wall, with a label giving the dates the machine was delivered and demolished. See my blog entry "Personal computers", at http://ljkrakauer.com/LJK/essays/pc.htm.

"Getting Started Computing at the Al Lab", 1982

MIT-AI is a modified PDP-10 KA computer which runs the powerful Incompatible Timesharing System (ITS). Along with the now-demolished PDP-6 computer, this old machine served as the primary computational resource for the first part of our history. Because of its old age, Al is not a very reliable computer, and most people do not store important files on it. MIT-OZ is a PDP-10 KL model B computer (a "PDP-20"). Acquired in June of 1982, this machine is intended to serve along side and eventually replace the services of the KA.

[larsbrinkhoff]

ALAN; ITSPAP 7

Sequence of machines: AI pdp6 DM pdp6 AI ka10 (first as attached processor; role switch) DM ka10 (pdp6 flushed) ML ka10 (first non-special-I/O-oriented system) MX kl10 (expand ML group to high performance) AI ks10 (begin renaissance) ...

Converted for KL in 1975(running in December). (Model A microcode only.)

Converted for KS in 1984.

[larsbrinkhoff]

AI KA10

Project MAC Progress Report IV, 1967:

It appears that at some time within the next year, it will be desirable to add a second processor to our central computer, the PDP-6. The initial selection of the PDP-6 computer was based in part on the fact that it was the only major-machine available that had built-in room for expansion, not only in input-output channels (which have grown steadily) but also in memory channels and in additional central-processor capacity. The manufacturer has continued to develop this concept and can offer a completely program-compatible second processor. This new processor, called the PDP-10, is about twice as fast as the present PDP-6, and if added with a small amount of its own very fast memory ought to triple the system's bulk computation rate.

RFC 254 from October 1971 says:

The MIT PDP-10(AI) system uses the ITS operating system and is similar to the MIT PDP-10(DMCG) system. At present the host is not connected to the ARPANET.

RFC 342 from 15 May 1972 adds AI as a Network User. In RFC 344 is had become a server.

https://github.com/PDP-10/its/pull/403#issuecomment-395183650 has the time AI started running TCP/IP:

MIT-AI started running ITS 1312 on December 31 at 21:03-EST

MSG: AI 1
Date: 05/02/83 13:51:15 From: CSTACY @ MIT-MC

The AI KA10 has been flushed, please update your programs.

Date: Sat, 8 Feb 86 03:15:22 EST From: "Pandora B. Berman" CENT%AI.AI.MIT.EDU@MC.LCS.MIT.EDU Subject: Reports of ITS death have been highly exaggerated To: macrakis@HARVARD.HARVARD.EDU, sollins@XX.LCS.MIT.EDU cc: KS-ITS%AI.AI.MIT.EDU@MC.LCS.MIT.EDU Message-ID: <[AI.AI.MIT.EDU].12905.860208.CENT>

the semi-original AI (that is, the KA-10 rather than the PDP-6) was flushed 3 years ago. the hardware was given to a bunch of hackers from Concourse (an MIT alternative undergrad program) -- after it walked across the street to Bldg. 20, the KA still ran, but had no memory, since its latest memory incarnation was all modified LispM Mem boards, which were given to needy Lispms when the Lab flushed the machine. i think the Concourse hackers have since had access to more of these Mem boards, but they have not (alas!) managed to bring the KA up in full-fledged fashion.

the AIKA went to the concourse computer club folks (an MIT student org.). we told them to tell us if they later decided they didn't want it any more. when peter lothberg came over here to pick up MC, his shipping container had plenty of room left, so we checked to see whether the CCC folks still had it/still wanted it/it still worked in any fashion. we were horrified to discover that Gill Pratt had just sold it to a scrap dealer for its materials, apparently to make room for some sort of solar-car project he was then involved in, without any notice to us whatsoever.

@hga, can you tell us more about the AI KA10, from the CCC perspective?

[larsbrinkhoff]

CENT; TAPES SAVE

Project MAC, whence descended the AI Lab and LCS, developed a highly unusual operating system called ITS (Incompatible Timesharing System). For many years it was the only, or the chief, OS used at the labs, so most of the seminal work done here was done on machines running ITS. Full ITS first ran on the AI PDP6, and was ported to the DM PDP6. Later, PDP10s became available, and the labs acquired some of the earliest ones -- the AI-KA10 (AI Lab's machine), the ML KA-10 (used by the MathLab, Theory of Computation, Automatic Programming, and certain other LCS gruops), and the DMS KA-10 (Dynamic Modeling systems, also used by certain other LCS groups); these replaced the PDP-6s, which were slowly phased out. The well-known MACSYMA program was started on the AI-KA, moved to the ML-KA when that machine case online and there underwent much development, and finally needed some space to really run. At that point the Macsyma Consortium was established and bought MC, the first KL-10A installed outside DEC.

[larsbrinkhoff]

The Australian FU KS10

Huw Davies:

Flinders University had a 36 bit system (I’m not sure if it was a KI or KL). I think they were running TOPS-20 on it. I see in the notes the they were running ITS on a KS

---

These Australians are really serious.

(1)How may I obtain a source tape of everything useful in ITS? Is this a reasonable request? I know they are already sexed up about bending a 1090's microcode, and they would like to know much more about ITS.

(2)These clowns really think they can do a VLSI pdp-10, and create a new market for a 10 in a terminal. I said "address space too small error" and they said "look at VAX: a 10 gives you about the same number of bits to address". I said "software unsupported error" and they said "good, at least the software has a fighting chance if DEC stays away from it". So... giving them the benefit of the doubt ... I still intend to send them what info they need, hoping always that someone else will attempt the project rather than my company.

Sure, we can send them a tape. We can probably even send them a 9-track tape. (Now that AI's tape drive is up. The last people we mailed an ITS tape had to find a 7-track drive!) There should be a "standard" ITS distribution tape set by the end of January or so, for the benefit of people with 2020s. That set would include, besides 2020 boot tapes, a tape or two of source code, and other things any ITS should have in its filesystem. We could just mail them that tape, and a description of ITS DUMP format (which is trivial).

[larsbrinkhoff]

MC KL10 (also MX)

Project MAC Progress Report XIII, page 69:

The Consortium purchased a KL-10 computer which is approximately five times as fast as our older KA-10 computer. A DEC KL-10 system was purchased for the Consortium largely from ARPA funds, and was delivered in July 1975. Guy L. Steele and David A. Moon rewrote the microcode for the machine so that it simulated the pager on the Mathlab KA-10 and our ITS operating system was in operation in November.

Date: 14 FEB 1976 0035-EST
From: MOON at MIT-MC

MIT-MC (the Macsyma Consortium KL10) is now on the Arpanet

https://github.com/PDP-10/its/pull/403#issuecomment-395183650 has the date from 1982 MC started running TCP/IP:

MIT-MC was doing TCP on December 19th.

Date: Thu, 8 Sep 88 21:39:25 EDT From: "Pandora B. Berman" CENT@AI.AI.MIT.EDU Subject: The end of the world as we used to know it To: (MSG ITS)@AI.AI.MIT.EDU, ARPANET-BBOARDS@AI.AI.MIT.EDU, INFO-ITS@AI.AI.MIT.EDU Message-ID: 439827.880908.CENT@AI.AI.MIT.EDU

"The time has come," said LCS, "MX at last must go. Its day has gone. We need that space Most urgently." And so Before we crate it, let us give A final cheerio.

Once there was a KL-10 called MIT-MC which belonged to the Macsyma Consortium. It provided Macsyma, the symbolic algebra system, to researchers all over the world, and mail gatewaying and mailing list support to a large fraction of the Arpanet. Things continued in this fashion from 1975 to 1983.

When the Macsyma Consortium dissolved in 1983, MC turned to providing cycles for MIT's Laboratory for Computer Science, and continued supporting much of the Arpanet's mail service. But the machine itself was growing old and cranky. In 1986, the mail services were moved to a smaller, more maintainable machine (a KS-10), and the name "MC" was moved with them. But the KL-10 continued to run under the new name "MX".

Now the end has come. MX was down cold for several months, and has only been revived recently to copy some old 7-track tapes. LCS can't keep MX any longer -- it needs the space for other purposes. So the KL is being sent to the Home for Aged But Beloved PDP-10s; a crack team of hardware hackers will arrive next week to dismantle it and take it back with them to Sweden.

Last seen online September 16th 1988, according to AI: BAWDEN; UPTIME DATA.

GLS wrote to ITS-LOVERS in 1996:

MC had special "ITS" microcode to emulate the SC pager ...

I was the one who wrote that microcode (summer of 1975). One hundred lousy lines of code and it took me three months to get it right! (Well, that included a few small mods to ITS.) That microcode also supported the JPC, with an improvement: as I recall, since there were lots of registers sitting around, it saved that last 16 PC's, not just one. That came in handy once or twice.

While I was at it, I found a way to speed up SOJx by one cycle; originally it was a cycle slower than AOJx and I found a way to make them dead even again. Since SOJx was a heavily-used loop control instruction group---more used than AOJx, actually---I thought this might be useful. DEC decided not to accept the modification on the grounds that it would be a maintenance headache (it involved having the central effective-address calculation do some extra work on the side with otherwise unused functional units, purely for the benefit of SOJx).

[larsbrinkhoff]

ML KS10

Date: Thu, 25 Jan 90 04:43:45 EST From: "Pandora B. Berman" CENT%AI.AI.MIT.EDU@MINTAKA.LCS.MIT.EDU Subject: adios to ML To: BUG-ITS%AI.AI.MIT.EDU@MINTAKA.LCS.MIT.EDU, POSTMASTER%AI.AI.MIT.EDU@MINTAKA.LCS.MIT.EDU, NGL%AI.AI.MIT.EDU@MINTAKA.LCS.MIT.EDU Message-ID: 693233.900125.CENT@AI.AI.MIT.EDU

alan has made an Executive Decision that trying to make ML work again, as well as keeping AI and MC in good shape, is Too Much Work For One Overloaded Grad Student However Encouraging His Friends Are. therefore, ML isn't coming up again. we did a full dump just before it came down in november, so all the files are available.

Last seen online January 1990, according to AI: BAWDEN; UPTIME DATA.

[larsbrinkhoff]

MC KS10

Note: it was called MX first, and then switched name with the KL10 MC.

Date: Fri, 28 Mar 86 09:02:47 EST From: Alan Bawden ALAN@AI.AI.MIT.EDU Subject: A new ITS is born To: INFO-ITS@AI.AI.MIT.EDU, KS-ITS@AI.AI.MIT.EDU, KARENS@AI.AI.MIT.EDU Message-ID: <[AI.AI.MIT.EDU].22057.860328.ALAN>

MIT-MX came up for the first time this morning. (You can supdup there right now, but you won't find much when you get there...) There are still some problems with the technology for creating new ITS systems from scratch, but it mostly works. Hopefully after doing the next two (ML and MD) things will be pretty smooth.

All kind of worms are crawling out of the woodwork because of various programs that -know- that all ITS machines are named "AI", "MC", "ML", or "DM". It should take another days hacking to stomp them all...

Date: Fri, 25 Apr 86 10:15:06 EST From: "Pandora B. Berman" CENT@AI.AI.MIT.EDU Subject: moving day To: KS-ITS@AI.AI.MIT.EDU, LAUREL@AI.AI.MIT.EDU, sollins@XX.LCS.MIT.EDU Message-ID: <[AI.AI.MIT.EDU].30845.860425.CENT>

MX and ML have undergone a mutual brain transplant. MX is now the KS between the LHDH bay and the 8N-HUBs, while ML is now the KS over beside MD, near the dover. MX is (still) running ITS; its LHDH doesn't work for some reason, but JTW et al are planning to work on this. ML and MD are still running Twenex, but that won't last.

[larsbrinkhoff]

AI KS10

Date: 27 June 1984 09:52-EDT From: J. Noel Chiappa Re: MIT-AI (sub 2) arrives! Wooooooo Shit! On the very loading dock at 545TS there is a monster moving van unloading a large collection of large boxes manifested to one M. Roylance, Esq. The THING seems to be here... They are sticking it inside Receiving since there doesn't seem to be a space ready for it on 9.

Date: Wed,17 Apr 85 17:19:39 EST From: Alan Bawden ALAN@MIT-MC Subject: AI ITS 1514 To: KS-ITS@MIT-MC Message-ID: <[MIT-MC].458514.850417.ALAN>

So today I typed ^Z on AI's system console and ITS started a job, loaded SYS:ATSIGN HACTRN which contained a program that opened the TTY: and did trivial terminal IO. Wow. Multi-processing and everything...

Date: Fri, 7 Feb 1986 09:33 EST From: @sraustein \ To: ks-its at AI.AI.MIT.EDU, cr-people at COMET.LCS.MIT.EDU Re: When we say reliable, we MEAN reliable

And it came to pass that on the seventh day of February 1986, AI the KS10, godchild of AI the KA10 and child of much sweat, fast talking, and cussing, did come up upon the ARPANET as host 10.2.0.6. And there was much rejoicing and jumping up and down and comments of "Yow, verily, we are winning".

And as we sat there, SUPDUPed in from various KL10s, running PEEK and marveling at the "(TCP)" indicators in the A display, we did see a small flurry of connections to the SMTP port. And we were somewhat bemused by this, since as far as the rest of the ARPANET could tell it had been 3 years since MIT-AI had last been up. And after a short time the flurry died down, and the entire net breathed a (quiet) sigh of relief that the two issues of SF-LOVERS and three Bandykin messages that had been caught in transit back when the KA was turned off had finally made it home to 10.2.0.6.

ARPANET. When it absolutely, positively MUST get there, even if we have to build a new CPU to deliver it to.

Earliest timestamps in KSHACK are from 1984-06-30.

[larsbrinkhoff]

DM KA10

AIM-238 says:

Actually the Project MAC Dynamic Modelling Group uses a non-paged early offshoot of ITS on their PDP-10.

RFC 164 from 19 May 1971 says DM was to have NCP by end of week. RFC 211 certainly lists DMCG among other hosts. RFC 148 seems to imply NCP was working to some extent even earlier.

@jh95468 confirmed that DM was the first ITS connected to Arpanet. (It's also strongly suggested by the host number; DM is host 1 on IMP 6. Multics is host 0, AI is 2 and ML is 3.)

An interview with Bob Metcalfe says:

I turned around and went down the street to Project MAC, now known as CSAIL, at MIT. I knew some people there and there were some openings. I said, “You have an IMP and you have a PDP-6,” and then a PDP-10 later. “I’d like to put those together, since I have just taken a couple of courses.” Al Vezza, who was part of J.C.R. Lickider’s lab said, “Okay, we’ll give you a job. Your job is to connect our IMP to our PDP-6

RFC 419 says:

on or about December 26, 1972, the system will cease operation for a period of at least 10 days and possibly as long as three weeks. During this period, we will be installing a paging system.

@jh95468 explained that the DM PDP-10 didn't have a disk initially. Hence there was no need for a pager.

He also revealed that there was backups being made, but the tapes were recycled.

MSG: END GAME
Date: 11/02/83 19:20:51 From: KMP @ MIT-MC Re: I see no DM here.

A few moments ago (7:05:50pm), the MIT-DMS machine was officially powered down for the last time. It is the second of four ITS machines to be retired, being survived by MIT-ML and MIT-MC.

"As the last syllable of your spell fades into silence, darkness envelops you, and the earth shakes briefly. Then all is quiet.

"You are standing at the top of a flight of stairs that lead down to a passage below. Dim light, as from torches, can be seen in the passage. Behind you the stairs lead into untouched rock."

The user who sent the original message merely scratched the surface of a sea of worms... -- Marc LeBrun

[larsbrinkhoff]

@bictorv When did UP first appear on the internet?

[bictorv]

UP probably appeared in 2004 (June?). Yikes, that's a long time ago.

[larsbrinkhoff]

And SV in 2001, according to Paul.

[aap]

I noticed the serial number of the DM PDP-6 is now given as 16. According to phil's list this is the Stanford machine!?! How can this be? AFAIK the Stanford 6 remained there for quite a long time.

[larsbrinkhoff]

Sorry @aap, I must have put number in the wrong place. Indeed we know very little about the DM PDP-6.

[aap]

Have you tried contacting anyone about this machine?

[larsbrinkhoff]

No.

[taa01776]

I don't recall that the DM PDP-6 was ever powered on during my tenure (starting in 1974), but it was there--as suggested above, just to the right of the KA's console. I can't be certain of this, but there is a distinct impression (someone must have said something) that the PDP-6 was serial number omega. Of course, there was a similar impression that AI's was serial number alpha, and I see from the list that it's 2. I should know the serial number for XX, because God knows I called field service for it often enough, but alas... DM had a lot of dead / sketchy hardware lying about: apart from the 6, there was some Fabritek memory that never really worked (probably started out on the 6). Main memory included an MD10, a couple of MA10s, ultimately an MF10, and an ARM-10 (256K in one box!). And there was the Evans & Sutherland LDS-1, which worked occasionally--whenever Ed Black went up and fixed some more cold solder joints that had failed.

[taa01776]

Immediately to the left of DM's processor was the SC-10 paging box, then some memory (I want to say that the MD-10 was there; certainly that's where the ARM-10 went when we got it), and the tape drive. Back row was mostly memory.

[aap]

I opened another issue https://github.com/PDP-10/-READ--THIS-/issues/5 to talk about the PDP-6 machines in general, unrelated to ITS. I really hope we can piece together more information.

[mgthompson]

ITS Hardware Note #1 details some of this. I will scan it this week.

On Sun, Jan 15, 2017 at 10:29 AM, Timothy Anderson <notifications@github.com

wrote:

Immediately to the left of DM's processor was the SC-10 paging box, then some memory (I want to say that the MD-10 was there; certainly that's where the ARM-10 went when we got it), and the tape drive. Back row was mostly memory.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/PDP-10/its/issues/181#issuecomment-272702251, or mute the thread https://github.com/notifications/unsubscribe-auth/AEFUHtuBfLjTYrU4zPWBb31C7oCa2XlVks5rSjtVgaJpZM4LHmwI .

-- Michael Thompson

[larsbrinkhoff]

Thank you, @taa01776. Some say there were 23 PDP-6s made, others 26.

Who do you think would know about the early times of the DM 6?

[aap]

I wrote an email to Gordon Bell to ask him if it maybe came from DEC. I doubt he knows this since he was at CMU at the time, but I still thought I'd ask.

EDIT: He doesn't know.

[aap]

Just now I realized that # 6 at Keydata was basically next door to the AI lab. Another probably candidate for DM, even though Gordon told me this some time ago:

Don't recall where the various serial numbers went, but Brookhaven Nat Lab and U of Western Australia Perth were the first few. Also KeyData and Rand. KeyData was returned. Am not sure when MIT got theirs or the Stanford AI Lab (SAIL).

Maybe returned and given to MIT?

[aap]

Greenblatt doesn't know either. He certainly remembers the machine, but thought it was new. He did say it wasn't the Keydata machine though because he knew that one quite well.

[larsbrinkhoff]

We got some new information from Lawrence Krakauer. This is a gem:

[36bit]

Both delivered and demolished on a Tuesday.

[larsbrinkhoff]

@philbudne Has Stanford CCRMA listed as a Foonly F4 running WAITS.

http://www.ultimate.com/phil/pdp10/10periphs

[larsbrinkhoff]

Actually, they had an F2 first, and later upgraded it to an F4.
https://ccrma.stanford.edu/~aj/archives/docs/all/646.pdf

[larsbrinkhoff]

@david-moon, 14 Mar 1985, on a dark period in ITS history:

With the recent episode of vandalism and destruction, there is only one running ITS machine in the world

That would be the KL MC, after the KAs had been shut down, and before the KS machines had came online.

(Later in the same message: "A perhaps little-known fact is that Symbolics still runs TENEX, on a Foonly F-2." Light bulb goes on... so that's why there are SCRC tapes with Foonex.)

[david-moon]

With the recent episode of vandalism and destruction, there is only one running ITS machine in the world

So I guess there are more now than there ever were before?

[mgthompson]

Theoretically I could install ITS on my KS10 and make it two.

Sent from my iPhone

On Jan 31, 2017, at 4:46 PM, david-moon notifications@github.com wrote:

With the recent episode of vandalism and destruction, there is only one running ITS machine in the world

So I guess there are more now than there ever were before? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.