Thoughts on DEC


I’m off this week, and while loafing around the house I took an hour or so to search the web about Digital Equipment Corporation (DEC), a former employer. That was a mistake, because it got me thinking about my past. This was fun for a while, and then it got sobering.

I worked for DEC from February 1978 until August 1996. During the glory years (before 1986) it was an adventure. We were on a world-changing mission. The code I wrote would live on and the work I did was so cool. Working for DEC was like working for Microsoft in the ’90s or ’00s, except that nobody hated your company. It was like working for Google except that nobody thought you were evil. It was like working for Amazon but pretty much everyone loved working there and it seemed like the world rooted for you to succeed. Most of all, everything we did clicked. There were some bad decisions, but the company always recovered and went on to greater glory.

I wrote CPU microcode for the VAX-11/750 and VAX 8600, worked on some cancelled projects that nobody will ever know about (ECL microprocessors, large ECL mainframes, a new RISC architecture), and my last project was working on Windows NT at DECwest.  In its day the VAX 8600 micropipeline was the coolest thing since sliced bread. How many people today know or care about it? Zip.

If you search for VAX-11/750 microcode, you’ll find a couple of tangential references. VAX 8600 microcode, ditto. A couple of papers, a memo or two. The spec I wrote about how SAFE would emulate a VAX, or my Argonaut specs or memos? Nothing. The cool things we did are lost to history.

I have a Venus Ibox microcode listing in my closet. It doesn’t exist anywhere on the web, AFAICT. Should I scan it and post it as a PDF file?  Who would give a rat’s ass?

There were intrigues swirling in DEC in the 70s and 80s: The politics surrounding the Minnow and Dolphin projects, the agony of the Jupiter and Jupiter II projects. Hardware alternatives that would have changed the arc of DEC’s history if they had been followed. Dolphin would have been a mainframe that ran PDP-10 and VAX instructions — 36 and 32 bit modes! How cool would that have been? How much more sensible than funding two separate development projects? How DEC shut down the PDP-10 business was incredibly stupid. (Yes, no, Yes kind of, maybe, I guess not.) Watching all that go down from nearby (I worked on Venus just down the hall from the Jupiter folks) was exciting.

Now? You can find some of the memos and specs online, but not everything. Plans for Jupiter multiprocessing? Just a couple of the memos. The really juicy memos about Jupiter’s performance problems are missing in action.

It’s sad. I thought our work would live on, if not forever, certainly more than a mere decade or two.

During my last job search, I met with one recruiter who started our conversation asking the usual questions, including where I first worked. I replied that I worked for DEC, blah blah blah. She replied, “Deck? What is deck?” I said that it was Digital Equipment Corporation. She never heard of the company.

As I read the few memos I found in my searching, I remembered most of the names I saw on the From, To, or CC lists. The recognition was a hammer. At one time those names were the anchors of my technical universe. A memo from XYZ was something you wanted to get your hands on. XYZ was talking!!  Now, who cares?

Sigh.

I’ve always been fascinated by the past and our relation to it. What went down before me, and before us. How people lived in times past. I don’t much care about how famous people lived or died. I’ve been far more interested in the common individual. How did people walk down the street and say, “Good morning,” in 1500? What fears did they have, how did they deal with their lifes’ issues? What did a factory worker think about in 1905? When I compare what I know about just one company to the available data, I’m amazed and saddened by what’s lost. There are a billion things about DEC that are lost to history. And that’s just one company that existed for 40 years. I’ll extrapolate and say we know less than 1% of 1% of 1% of what life was like in the past. We have historical records, newspapers, paintings, poems, and photographs, so we think we know what went down. We don’t. The overwhelming majority of human information has been irretrievably lost.

And I’m not talking about thousands of years ago. I’m talking about just 100 years ago.

So back to my work history. Does anyone care about Venus microcode, or Comet microcode, or SAFE memos? How about memos that argued for continuing the Dolphin project? No. It was once the center of my universe, and now it’s crap.

Do other people think like this or is it just me?

17 comments
  1. I am a few years younger than you are, I think, but I remember the VAX 11/750. I used a cluster of these in college, from 1985-1989. I am more familiar than most with the VAX instruction set architecture, or at least I was, because one of my class projects was to write an assembler, in assembly language. It was incomplete but it worked, and I remember with glee how a printout of the source code would stretch down the hall of the computer center — while debugging this code my fellow students and I would crawl up and down the listings.

    The writing was on the wall for minicomputers, though, and I also worked on a mini-operating system for the 68000, and C programs for the early Macintosh systems.

    Just last fall I taught a class on system software, including the basics of assembly language programming. I asked the students to write very simple assembly language code for the 6502, running in emulation. That did not go very well. But I did also mention the VAX, as an example of a system that had registers that could be dedicated to managing memory for programs compiled from higher-level languages: the argument pointer, frame pointer, and stack pointer. It was, as Wikipedia puts it, the “quintessential CISC ISA.” In my opinion, a great architecture. I’m afraid the microcode is probably a bit beyond me… but I’m glad it was there!

  2. Jon Jacky said:

    Computing culture has a weird amnesia about its own recent past.
    There are several problems with this, and your article made me aware
    of one more — it can cause people to devalue their own experience and
    their own past.

    Your past work at DEC (and elsewhere) is not crap – you should be
    proud of it! Beyond any value it might have had at the time, that
    experience also informs everything you have done since. Likewise, the
    collective products and experience of you and your colleagues is what
    made our present technology possible — whether people realize it or
    not.

    PS – Other people do think like this. It is not true that nobody
    cares about the technologies and projects that are the recent
    ancestors of today’s. There are historians who seek to recover them
    and trace their influence, and engineers (and others) who try to
    restore or recreate them. What motivates them (in part) is the same
    thing you mentioned: they don’t want our past to be irretrievably
    lost.

  3. tom biggs said:

    I learned to program on PDP-11s, working at a company that was a reseller of DEC equipment – we wrote the software and sold the hardware and software as a package.

    I can still remember some of the octal opcodes for the PDP-11! What a great architecture. I moved from FORTRAN and BASIC to assembler, and I remember thinking “why do people say that assembler is so difficult?” I found out later when I moved to x86. Ugh.

    DEC made our company’s fortune and truly, everyone was fond of DEC.

    But, nothing lasts. My brother is in construction, and we can drive around his county and see houses he built 20 or 30 years ago. I don’t have that. Everything I ever worked on was obsoleted, cancelled, or dropped when the company changed direction or went out of business. 33 years in the business and it’s all gone, all that work no more lasting than peeing in a river. I’ve grown philosophical about it, though. As long as I’m having fun and learning right now, it doesn’t really matter where it ends up.

  4. So the bulk of the iceberg that was DEC is below the waterline of the internet. Just think how much history is below the waterline of the printed page. 😛 – I know that you alluded to this, but i wanted to drive it home a bit further.

    So there are a wealth of vastly more banal efforts that are vastly better documented solely due to the fact that they started after 1997, sux, but oh well.

    But here’s the deal, DEC is greatly represented, because the echoes of DEC reverberate like the big bang of modern computing; DEC’s history is the cosmic background radiation thereof.

    When I read your post today i immediately thought of the MOTO 6800 architecture that is what apple’s empire (and sun’s, and some others that escape me) was founded on. Totally cribbed from PDP instruction set, think i even read that in the introduction in the data book – note the transposition of the 6 and the 8 eh!. x86 is so ugly by comparison. It was interesting to me that the first poster brought up moto as well.

    And many of the other things that you mention died at DEC but where reborn in other places when all of you refugees bailed or where blown out of the building. Bet you can find bits of

    We have desktops->laptops->notepads->google-glasses->implanted-wetware due to ken olsen and y’all. tnx!

  5. I loved DEC documentation. It saved me when I was working on a project in the early 90’s 4000 miles from home. You should be proud to have worked there! I still aspire to produce such high quality work (hardware, software and documentation) as I have seen coming from DEC.

  6. Eric Fiterman said:

    I think you’re right on about history and how much actual knowledge is lost through the generations. Until recently, I thought my parents’ generation would be the last to know that humans walked on the moon. Imagine if we didn’t go back into space… probably in less than 100 years people would think that was impossible fiction. Maybe there’s a reason the ancients built pyramids and megalithic structures – that’s certainly a more permanent way to deliver a lasting message (although no one may be able to interpret it).

    In the end, I think this is the nature of technology and the age we live in. That’s why the name of the game is to do meaningful work and have fun doing it (which it sounds like you do).

  7. tom miller said:

    “Should I scan it and post it as a PDF file? Who would give a rat’s ass?”
    I would.

    • John said:

      Hmm. Well, I’ll see about doing it, then.

    • John said:

      It’s been scanned. Picking up the file tomorrow.

      • tom miller said:

        Looking forward to seeing it!

    • John said:

      Here you go, Tom.

  8. Stevie Wonders said:

    I did my first assembly on PDP-11, then 68000, wonderful. Then encountered x86, gag. Computer tech is so strange in rejecting prior knowledge and experience. Does anyone younger than 45 realize the clever techniques to maximize utility of early microcomputers are same as needed for today’s low cost microcontrollers? Or the struggles to adopt features like multitasking, priority scheduling, and protected virtual memory to PCs were already well understood in the mainframe world by the late 1960’s? Yet Microsoft, Intel, et al acted like they pioneered this stuff. Same with Sun’s Java virtual machine, anyone recall Pascal pcode? Or that Postscript is a derivation of Forth, developed in 1968? The more I study computer history, the more I see how little that seems new truly is.

  9. Johnny Billquist said:

    As a blast from the past… I was searching for VAX 8600 documentation online for some other strange reason and hit upon this page and your scan of the IBOX microcode. What a pleasant surprise! Thanks a lot for that. Just so you know, there are people who really appreciate this kind of information.
    I can confirm that no newer version than 3.73 of the IBOX was done. I actually still have a running 8650, and I checked the version on that machine.

    Any other documents you have, I can assure you that there are people who would be most interested in getting any copies. Just contact me if you want.

    • John said:

      I have a few more things. I’ll copy them and post them here!

  10. Our careers overlapped fairly closely: I joined in late 1977 and stayed until the middle of 1999 (when DEC was already Compaq-ted). And yes, most of what we worked on is forgotten, although there are a surprising number of companies that still run VMS on VAXes and Alphas. VMS is once again under active development, at VMS Software Inc. in scenic Bolton, MA.

    Still, I regard the work that was done as still relevant to any engineer who wants to work “down stack” rather than up in the ether. There is continuing interest in old hardware and how it works. The battles of the 80s – macro-pipelining vs short tick, CISC vs RISC – proved surprisingly relevant this century.

    So who would like the Comet (750) microcode? I would, for one, And a copy of the RH750 spec, if you happen to have it lying around. 😉

    • John said:

      It’s funny what I have discarded (Venus Ebox microcode, lots of Venus and Dolphin memoranda) and kept.

      I’ve got a Comet microcode listing dated 1991-01-19. Complete. I’ll take a trip to a copying place this week and put someone’s kid through grad school.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: