If It Ain't Broke, Don't Fix It: Ancient Computers in Use Today
pcworld.com
pcworld.com
Here's a sample of a great mod: http://www.youtube.com/watch?v=dZWgXiVJbpI - albeit from a lame demo. There are thousands more available at aminet.net.
I've tried replacing the machine with emulation, but there is something comforting in seeing the old girl in the corner serving as well as she ever did. Perhaps devices living beyond their natural lifespans comforts man's worries over his own perishable nature.
They do make great coding music. Not coincidentally, .mods resemble code, being in a way code themselves, with tightly structured and repetitive patterns but also with their own creativity flowing over the top.
Necros / Isotoxin
He kept one in his detached garage that would load the initial bootstrap and then prompt the console for the operating system, which he didn't have. From that point on it patiently waited for a response and was used to heat the garage.
$
But the Operators never arrive.
boot> _
The developers should have put a countdown timer in there. Each hour, add a question mark and change it to:
boot?> _ boot??> _ boot???> _ hello? boot?> _
[1] http://en.wikipedia.org/wiki/Harwell_computer
[2] http://www.youtube.com/watch?v=vVgc8ksstyg (the video doesn't really do it justice)
There's a button on a long wire which you could click to advance the current program a single step. You could watch the RAM spin and flicker and the relays(?) 'clack' through the computations :-D
It's a great museum because of how massively hands on it is. I can't think of many other places that have that high percentage of exhibits which are there for kids (and adults) to play on.
It occurs to me--is there some sort of financial instrument a company could buy (or sell) which would incentivize them to take a longer-term view of their own performance?
The one thing that comes to mind is Enron's scheme of selling debt backed by the integral of future projected profits from an energy contract, and then writing that debt-asset down as current-quarter earnings. It did overvalue their stock, but it also locked them into a model where contracts with declining long-term outlooks would force a loss, rather than just a declining gain.
That doesn't seem like a glowing recommendation, I admit--but an overvalue of the stock in this case could actually be the "economically correct" value; it could just be incorporating the company's potential upside for being forced to take a longer-term view, and rewarding the participants for taking on increased downside risk.
The stock price, in an efficient market (that is, one that understands what it's investing in, unlike what happened with Enron) would also be offset with lower demand due to said increased risk--but a company could still come out slightly better off for establishing these instruments. It just comes down to whether other short-term investors actually value a company that "goes long" more than a regular quarter-to-quarter-earnings one.
I think restricted stock is basically that.
ESOP. When the stockholders are the employees, quarterly earnings take a back seat to making sure the company will still be successful 30 years from now when they retire.
I think it's really cool that some of these machines are still running just fine after 25-50+ years, and while I agree with the sentiment, it really scares me as a user, and would even more if I were a business owner. I couldn't even make it past the first section.
We spend all this time talking about data portability, using open, non-proprietary formats, and redundant backups. A fire could wipe out everything, as it would be a huge undertaking to either reproduce a copy for offsite archival (which is already likely the case). What happens when the machine breaks beyond repair and a replacement can't be found?
I'm all for repurposing old hardware, extending the useful life of machinery and not upgrading just because. But modern computing systems have huge advantages including physical data size, redundancy and search, among others. Not to mention that even if the Sparkles decided to move to a modern system, the data would have to be transferred manually.
The chemical company in Texas is a prime example: "They use it because its a known entity."
Wow.
I would assume that this is the single reason outdated systems continue to exist. People who are using the old systems and feel confident with them. If you want, you can find for any legacy system a person who would be willing to take the risk to replace it. But that would also mean losing control for the existing maintainers...
I wonder what the risk/time graph looks like for that compared to say, the MTBF of a modern hard drive.
I got a VAX post-Y2K and ran VMS and NetBSD just fine.
The OSes were, but what about the apps in common use?
It's the same thing on ancient microcomputers: As far as I know, the C64's ROM-based OS doesn't care about the date at all, but various application software likely did, and would probably get very confused at two-digit years which began with a first digit less than 7.
Work was in progress to create the next generation using more modern hardware and software, but the old system had to be kept up and enhanced until the new version was finished and deployed to all of the remote sites. The biggest impetus for change was the difficulty in finding replacement parts for such obsolete hardware.
Unfortunately, an emulator just wasn't an option. Being a real time system, the timing constraints were too strict. Think of the old DOS games that became insanely fast on more modern hardware, but with more significant consequences. For better or worse, I saw sections of code where there would be, say 16 no-op instructions, and a comment like "The following 16 no-ops give the bit-sync just enough time to finish adjusting before the real signal arrives". Tracking down all the timing dependencies and verifying they still worked on an emulator would have been a nightmare, and probably more difficult than creating a new system from scratch on modern hardware.
All of the dev tools were hosted on a VAX system. Fortunately, this one was able to use an emulator, and it actually ran significantly faster than the real system, and increased compile time dramatically. They still kept the real VAX around in the data center, but at some point the replacement parts and system maintenance costs became too expensive.
The two guys in the room (I'm guessing that room is always manned, even during a refit) immediately understood our concern. "No no," they said, "that's not the missile computer. That's the missile computer." They pointed behind us at a giant grey metal cabinet. I imagine the technology was from around 1980, but from the look of it I wouldn't have been surprised to find vacuum tubes inside.
We were all quite relieved to find that nuclear missiles are controlled by an old computer that presumably doesn't receive security bug patches on a weekly basis.
That missile control computer better not be on any network..
If it works, it works. The cost to "upgrading" something like this is staggeringly high when you factor in the likelihood that the change is going to break something.
About the only change is I no longer use a CD player, cassette player, etc., but drive it from an Audiotron box that sucks the tunes off of my computer. The Audiotron is > 10 years old, and I still haven't found anything better.
(The Carver amp on-off switch did wear out last year, but I just plugged it into a bus strip to use as the switch.)
That won't last long, 386 support was removed from mainline kernel.
You should definitely do a write up on how you're getting 8 hours out of a PC-UM10. Didn't they get like 3 hours under Windows? I could use a simple Linux laptop, even if it's terminal only.
Edit: No, apparently it was switched off in 2010, I remember talking to somebody from their IT department back in 2008 when that computer was functioning, didn't know that they actually shut it down now.
A couple of years ago my father and I scrapped the last of the Rainbows (not rare enough for a museum to be interested) and a Micro PDP 11/23 (no interest from museums or UK PDP enthusiasts, and too hard to ship internationally). It was a sad day.
(I still have several boards from a DG computer on my office wall. Geek porn...)
I agree that its a great book, these days it would be much less dramatic with everything running on simulators prior to chip tapeout.
My Mom worked at Data General and she used to take home some transparent masks for chips that were "corporate trash". She made art out of them and they're still hanging in her house.
Same here, pretty much. That was a very influential book for me. I went back and read it again a couple of years ago, and it was still fascinating. Then I found out a guy at our local Hackerspace was there during some of that time and knows some of the folks mentioned in the book. That made it all seem even more real to me... quite an interesting story in any case.
We did look at some cheaper options, but it's interesting that the manufacture of electronic systems doesn't really have anything cheap in the space. Properly managing nested BOMs of an arbitrary depth? Pretty much always a high pricetag or something you have to hack yourself...
IE6 just becomes the shell to the webapp instead of a browser in its own right. MS probably even has a way to install multiple versions of IE concurrently so you could run IE6 + "modern IE".
On Windows I think the official way is runnng it in a vm.
You can also make a nice living writing xBase (dBase/FoxPro/Clipper) code. Millions of lines of xbase still happily running along.
I guess there are still a few people using abacuses, if that counts.
My partner worked for an insurance company and a lot of their infrastructure is from the late 80's so is the software and it's always worked for them. Their reason for not switching to modern equipment is that it is more reliable and their business centers around data. If the software or data were to fail them in anyway that's potentially a lot of money or worse credibility and trust lost. I think they saw it as being cheaper to maintain the current system and I can't blame them, a lot of banks are the same too.
Is built better, maybe. But "works better"? It would take that thing 20 minutes just to load Ghostbusters. That in no way could be considered working better than today's computers/consoles.
The solution to this is simple: Pick some arbitrary point in the past, and declare that every feature which became common after that point is 'bloat' that 'nobody really needs'. Obviously, a good program doesn't have bloat, and a good system can't be expected to run well unless it is given good programs, so all good systems run well, which means better than anything that will run bloat.
Outside of my home, I saw an ancient DEC PDP/11 still in use at a newspaper in North Carolina as late as 1999 when I worked there. They also had an old IBM S/36 which only got replaced with a (then) modern AS/400 box about 1999. One of the machines they had in there (and I'm not even sure which one it was) used those old drum-based disk drives, with the big drum with the spinnable handle on top, that weigh about 20 lbs each, and old a whopping 50MB of data.
And I think that's kinda the point about a lot of these kinds of stories. Everybody knows that there is a better solution, but inertia, lack of resources, and or risk related to the new solution, keep people from doing anything about it.
So, one day sooner or later (probably sooner) the power supply in that thing will crap out, and I'll bite the bullet and reconfigure my network to use my wireless router to do all that stuff. Unless... I think I may have another one of those power supplies in a box in the closet somewhere....
What pulled the trigger on this project was that the factory's management noticed that the last guy knowing the machine and the whole process design was retiring. I changed jobs in the meantime and I bet the thing still runs to this day: the planned migration will probably take years but at least enough knowledge and data has been passed on for that project to complete.
Surprisingly, a new one sells for double what I paid for.
I have an emulator on my phone but the tactile response of the keys is awful and the display and keyboard are tiny compared to the "real" thing.
Its just too easy to use. Latency of the calc on my phone is very high compared to a fraction of a second. If my desktop is up and running and logged in and a webbrowser is running (lots of ifs) then google and wolfram alpha are starting to eat into the 48's territory. But it still gets plenty of use. Batteries last about a year.
I also have a 32Sii at the workbench vaguely 90s vintage.
In 1985, I was preparing for the matriculation exam (at end of high school). I was worried that my Sharp EL-506 battery would run out in the middle of this important exam, because it was already 3 years old. However, I did not change it. It lasted the matriculation exam. It lasted also my MSc studies and army, 9 years altogether. And beginning of my working career. And it actually still works, though nowadays I don't need it so it's just at the bottom of the drawer. But it is OK, 31 years after buying it, with original batteries and some significant use during high school and uni. I just don't understand.
TI has a pretty good gig going since they're selling a "calculator" and not a "computer". "Calulators" don't follow the same market dynamics as "computers", so as long as everyone in that market plays along and competes on features (or no one adds features) they should be able to maintain their margins ad infinitum
I certainly would take my business elsewhere if I discovered the company was doing their accounting on a punch card machine because the 60-year-old white guys were too afraid to learn how to use windows.
With typical Windows PC would come occasional hardware failures, crashes, data losses and viruses which all are serious trouble if you don't have somebody knowledgeable on site to mitigate and fix them.
But seeing it in use just screams "I fear change" to me. In what other ways are these guys in the dark ages?
* All their "software" (ha!) is stored in physical form, on a shelf, as a board full of wires--hard to mistakenly delete it or lose the registration key.
* Their code has 60 years of testing behind it
* Cards provide a physical record while also being machine-readable and, as I pointed out elsewhere, if a card gets mangled, well, you just sit down and type up a new copy
* They'll never lose your credit card details to hackers
* They're comfortable with the technology and have a very clear process. You'll never call them to hear, "Oh, well, they just updated the system and now I can't find the link to our invoice archives"
With the average computer level skill in the society, the likely alternative is your order information being stored in an Excel file in a Dropbox account with password "files123." I'll take the punch cards.
I love my Tektronix 2430A. I have some of the Tek 160 modules for the modular oscilloscope from the 50s, although I haven't really tried that. If you saw the guts, it's a point-to-point wiring masterpiece.
I find this truly fascinating.
Eight
Megs
And
Constantly
Swapping.Knowing you can update or change your system implies nothing or no one is really indispensable. I would much prefer something like that.
"A slow sort of country!" said the Queen. "Now, here, you see, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that!"
- Alice in Wonderland
I'd be more interested to hear about computers that are in use being actively changed, with new software coming out on them to support whatever unique need they have.
That or highly specialized one-trick-ponies that cannot change due to their complex nature.
The military mass producing 100,000 <insert thing> from the 70's with computers on them, or a person maintaining their BBS doesn't really feel the same.
The allen bradley/rockwell 500 series PLCs have been in continuous production since the 90s, lets say the 5/01 has been made from '95 (although probably earlier?) till today as far as I know, so that's 18 continuous years of production. They're pretty simple but sometimes that's all you need.
If you allow a compatible family orientation, the first vax rolled off the assembly line in '77 and the last in '05 so thats a good 28 years of continuous production. Of course thats a trap, I would imagine "PC/XT compatible" boards will be made basically eternally for embedded work, etc.
AmigaOS is still being commercially developed, though it 'only' dates back to 1985: http://hyperion-entertainment.biz/
And there's new hardware for it: http://www.a-eon.com/ http://acube-systems.biz/
The market is tiny, and mostly hobbyist focused, but there is even the occasional commercial software release, now often targetting AROS and MorphOS (AmigaOS inspired OS's) too. For example:
http://en.wikipedia.org/wiki/Hollywood_(programming_language...
The Amiga community is interesting because there is a distinct split between different factions, with some insisting on only supporting "classics" (the M68k machines) and sometimes reimplementations (there's a series of FPGA based projects), some only interested in the current generation PPC systems (from AEon and ACube), as well as AROS and MorphOS camps, and those only interested in emulation.
Of course a lot of people couldn't care less about the various splits and just want to get on with things, but this is a community where you will find people actively using anything from 7.16MHz A500's to high end PC's to run OS's that are widely source compatible and that either directly runs or have some level of integrated emulation capability for old M68k apps, and where a lot of the community still run and/or tinker with software that was released back in 1985 (e.g. there are people still tweaking the original roms to cut more cycles off the odd system call...).
That system apparently works only 60 percent of the time.
Indeed. If it's broken 40% of the time, don't fix it, replace it.
You probably don't even support IE8 anymore. :)