I worked for Sperry Univac 1974-79, in Halifax, Montreal and Calgary. I was an "SA", Systems Analyst at the service of the Sales team. It was a lot of fun. The Univac salesmen were the cowboys that didn't fit in at IBM. When preparing benchmarks, money was no object, we had lavish expense accounts. In the Oil Patch I saw $100K deals signed during coffee break.
One of Univac's problems was the proliferation of operating systems for the different incompatible architectures. There was Exec 8 for the premier 1100 series (36-bit); OS/4, OS/3, OS/7 and later VS/9 (formerly RCA's TSOS then VMOS) for the 9000 series (32-bit); also the 418 and 494 real-time OS'es (18-bit words). Then there was the CADE 1900. All written in Assembler of course. We even had Varian in the branch, with salesmen from the different product lines competing for business.
All this duplication resulted in overhead and squandering of programmer resources.
After the Burroughs merger, the joke was that UNISYS stood for "Univac is Still Your Supplier".
I spent seven years doing maintenance programming for EXEC 8, the operating system kernel for the UNIVAC 1108. It was first demoed in 1967, and was way ahead of its time. By about 1972, it was running reasonably well. This OS had threads, symmetrical multiprocessing, and async I/O. Written entirely in assembler, it was not fun to work on. When it crashed, a dump was taken to drum, and the dump was then printed, producing a stack of paper about two inches thick, mostly just pages of octal numbers. Analysis involves pencils and colored highlighters, tracking pointers through memory.
Here's the manual.[1]
Amazingly, the descendant of EXEC 8, OS/2200, is still a maintained product, over half a century later.[2] There's even a roadmap out to 2033.
I was curious about what OS/2200 was in the Unisys lineup. It didn’t fit with MCP and what I understood about it. At some point you could download en emulator that ran on Windows.
Ah, the 418! That's something I haven't heard about in a while.
My dad worked for Sperry Univac. He had a laminated list of 418 assembler instructions, with assembler mnemonics, and time of execution. I seem to recall 4 microseconds for addition and 6 for multiplication, but it's been a while since I saw it...
I worked for the minicomputer division(may be using the wrong term - it was a long time ago) from 80-83 as an pre & post sales support - likely equivalent to the "SA" on the mainframe side. Within a week of starting they sent me off to Irvine CA for a two week course on the OS internals. Somewhere I believe I have a book that is a printout of OS source. (all V77 assembler).
I have some great memories - first time seeing whales just off Newport Beach stands out.
The article shows up here fairly regularly. I mainly bring it up as the computer was developed by Sperry, and it really is a good read about an early tech project.
The founder of Sperry Corp is this guy https://en.wikipedia.org/wiki/Elmer_Ambrose_Sperry who has an incredible dossier of inventions to his name and his son Lawerence https://en.wikipedia.org/wiki/Lawrence_Sperry was an aviation pioneer and inventor who invented the first guided missile, autopilot, and much more, as well as being accredited to starting the "Mile High Club". I'm reading the Elmer Sperry biography currently and its incredible how many projects they worked on in such a short span. Makes you question how productive we are in the computer age...
> Makes you question how productive we are in the computer age...
The difference is that modern society isn’t optimized for productivity, it’s optimized for consumption and attention. 30 years ago you could go days without seeing ads, now companies send notifications to the nuisance devices in your pocket and on your wrist. It’s no wonder we have a harder time focusing and accomplishing things. Even being aware of the problem isn’t enough to fully protect one’s attention and intention.
> 30 years ago you could go days without seeing ads
If anything, it is easier to avoid advertising today than it was 30 years ago. You can block notifications, spam filters are significantly better, content filters at the web browser and DNS level are available. On top of that, certain nuisance devices are possible to avoid. For example: you can still buy traditional watches.
Contrast that to 30 years ago. Nearly all radio stations, television stations, news papers, and periodicals included a lot of advertising. Telemarketers were certainly a thing. Advertising in public spaces was certainly a thing. You could consciously avoid most of those things, but you would also be isolating yourself in the process.
Perhaps you could make that argument if you were talking about 100 years ago, of people who lived in rural areas, and had very little contact outside of their community. Even then, you are talking about a form of isolation. The only difference is that type of isolation was much more common.
It's optimized for profit. Ideally these would be one in the same. In any case our productivity is easily an order of magnitude greater today than it was 100 years ago. The lack of technical progress into our legislatures and courts even though we currently poses the ability to afford the same order of magnitude improvements is a curiosity and implies there's a form of corruption that's taken root. This seems to allow non productive companies to command most of the profits.
> a harder time focusing and accomplishing things
Relative to self reports? The grass was probably greener and the schools were definitely better too. You might as well note that jobs have not gotten any more complex nor opportunities any wealthier in that time. People still flip burgers and sling french fries for minimum wage. They used to goof off out back now they troll the internet in their off moments.
> isn’t enough to fully protect one’s attention and intention.
Outer space is super exciting. Until people realize it takes months to make it even just to Mars. Our attention span was never that great to begin with.
It wasn't as corrupt. Anti trust laws and laws separating banks from investment firms were active and enforced. Our legislature sure hasn't dwindled in productivity.
It wasn’t as perfectly tuned by machines of infinite knowledge and calibrated individually for every person over and over until they can no longer do anything other than generate a profit for the advertisers.
Some exaggeration above, but less than would be fair.
It wasn't technologically possible to have such a tight feedback loop 30 years ago. We realized the Human Instrumentality Project around 2013 and now every-goddamn-thing spies on us at all times so markets can react in real-time like the ultimate jealous controlling spouse — “Where are you going? Who else will be there? What are you talking about?”
As an additional burden, we've also seen the ascension of "bullshit jobs" - rather than increased productivity bringing 15-hour work weeks, our society developed pointless, if not net-negative-value, jobs, which consume countless hours of the lives of people.
One possible reason is many low hanging fruits, after WW2 with all that cheap industrial base and new technologies emerging from the war and almost no regulation with cold-war budgets.
You could use the tech and build it fast.
Check the "Secret history of Silicon Valley" by Steve Blank:
Starting from a blank sheet of paper, it is easy to make leaps and bounds level of achievements. After something matures, the achievements become more incremental and looks like the tech is stale.
Companies like Sperry Rand and Honeywell are like that. They don't put their logos on the outside boxes, but their logos are inside just about everything.
I've been familiar with the Sperry logo for about 50 years, but only because dad had bought a New Holland combine in late 1969 that sported the Sperry New Holland logo. It wasn't until more recent times that I learned the company was a conglomerate engaged in diversification but primarily known for large computers.
New Holland went through a few iterations until being purchased by the Fiat group. It is a part of CNH with its sister company of Case IH.
Please note that the "solid state computer" wasn't fully transistorized[1]. Much of the logic was magnetic logic[2]. (Which is solid state, but not semiconducting)
This happened at the time when Germanium transistors were hand made, and quite unreliable, less reliable that even vacuum tubes. Magnetic core was much more reliable than either.
More context: I help a friend who has been repairing stuff since the 1950s. I've learned from him, and from the germanium transistors we encounter (after decades in light service) that they tend to "go soft" with use, as do vacuum tubes. You end up replacing them with "new" germanium transistors, or re-bias the circuit to work with silicon transistor replacements that have no such issues.
Some great 1955 articles by aerospace journalist Ansel Talbert cited & linked to from archives in it.
Warning, the overall substack series is on UAP, a topic which HNers often find uncomfortable. So if the UAP theme offends sensibilities, don’t click through.
“The data processing system selected for the Apollo remote sites is being manufactured and assembled by the UMVAC Military System, Division of the Sperry Rand Corporation, located in St. Paul, Minnesota. The computer is identified as the UNIVAC 642 B Modified and has been designed to meet military specification. There will be two identical computing subsystems installed on each of the sites of the Apollo Tracking Network. These subsystems are identical in every respect with the exception of the mission requirements which will be assigned to each subsystem. One computer subsystem will be used for the processing of telemetry data and will also provide a command processing back-up capability. The second computer subsystem will be used for the processing of command data and will also provide a telemetry processing back-up capability. The purpose of the back-up capability is to provide continuous operation for the remote site computing requirements should either computer malfunction during a critical period of the mission.”
>What little hardware R&D in which the company was still engaged was split between AI focused LISP machines for which there was no immediate market, and mainframes that were a rapidly shrinking market
Sounds familiar, eliminate "LISP machines" I wonder if history is repeating itself today for many companies.
I fondly recall being a site operator and programmer on a Univac 1100/62. Remote sync terminals (UTS 40s?) and UTS 4040 cluster controllers. Fun times.
One of Univac's problems was the proliferation of operating systems for the different incompatible architectures. There was Exec 8 for the premier 1100 series (36-bit); OS/4, OS/3, OS/7 and later VS/9 (formerly RCA's TSOS then VMOS) for the 9000 series (32-bit); also the 418 and 494 real-time OS'es (18-bit words). Then there was the CADE 1900. All written in Assembler of course. We even had Varian in the branch, with salesmen from the different product lines competing for business.
All this duplication resulted in overhead and squandering of programmer resources.
After the Burroughs merger, the joke was that UNISYS stood for "Univac is Still Your Supplier".