Retrotechtacular: Amiga Pips The PC For Mission-Critical Computing At NASA

An Amiga computer at NASA

In 1986, a group of NASA engineers faced a difficult choice in solving their data processing woes: continue tolerating the poor performance of PC architecture, or pony up the cash for exotic workstations. It turns out that the Commodore Amiga was an intriguing third choice, except for the fact that, paradoxically, it didn’t cost enough. Oh, and Apple wanted nothing to do with any of it.

Steeped in history, NASA’s Hangar AE is a hub for launch vehicle telemetry and other mission communications, primarily during prelaunch phases for launches at Cape Canaveral. Throughout the late 20th century, Hangar AE supported NASA launch vehicles in all shapes and sizes, from the Atlas-Centaur evolutions to the mighty Titan family. It even supported user data from the Space Shuttle program. Telemetry from these missions was processed at Hangar AE before being sent out to other NASA boffins, and even transmitted worldwide to other participating space agencies.

Coming down from decades of astronomical levels of funding, the 1980s was all about tightening the belt, and NASA needed budget solutions that didn’t skimp on mission safety. The Commodore Amiga turned out to be the right choice for processing launch vehicle telemetry. And so it was still, when cameras from the Amiga Atlanta group were granted permission to film inside Hangar AE.

The video below was filmed in 1998, over a decade after the first Amiga computers were installed at Hangar AE. It’s fascinating (and unsurprising) to hear that the Apple Macintosh was the first choice of computing hardware. Being a closed system, however, engineers couldn’t access the Macintosh at the level required, and weren’t able to develop the custom hardware that was needed to support their operations. In contrast, Commodore were more than willing to send NASA an enormous stack of documentation to help them out. Nice of them!

Gary Jones, then Principal Systems Engineer for NASA, goes on to say that the Amiga was an unpopular choice for his employer. “They want us to buy PCs and run Windows 95 and NT. We keep trying to tell them its not fast enough so they tell us to buy DEC Alphas. We tell them its too expensive. They don’t like the Amiga; it doesn’t cost enough.” Yikes.

The video takes place during STS-89 and its mission to the Mir space station. It appears that some of it has been lost to time, however an old blog post fills in some knowledge gaps, it’s well worth a look and is retro-tastic in its own right. Reports indicate that these machines were in use as late as 2006, and one was actually for sale not too long ago.

So the jingle goes, “only Amiga makes it possible”.

[Thanks to JohnU for this great retro computing tip!]

[Pictures from Amiga Atlanta/Mike Ellenberg]

46 thoughts on “Retrotechtacular: Amiga Pips The PC For Mission-Critical Computing At NASA

  1. “you need an operating system that’s very efficient very fast and the Amiga has that it’s got very little overhead that’s what makes it nice we don’t load down the sister running the overhead we can just go ahead and process the data”
    On hearing that my first thought was that is the total opposite of MS windows bloatware.

  2. Boots faster from a floppy drive than any Widows did from a hard drive until Microsoft decided to hibernate instead of shutting down/rebooting.

    AmigaOS 3.2 on just about any hard drive stil boots faster than my Win10 wakes up from hibernation on an NVME.

  3. I was trying to work out what maths the Amigas 1000’s were doing solving up to fifth order polynomials.
    Then I looked at the display and saw temperature and instantly thought of theromcouples, they have an equation of the form:
    T = a0 + a1 * X + a2 * X^2 + a3 * X^3 + a4 * X^4 + a5 * X^5 + a6 * X^6 + … an * X^n
    Where X is in Volts and T is in °C
    An iron(+) constantan***(-) (Type J) thermocouple junction would be a 5th order polynomial

    So they were probably solving equations like:
    T = a0 + X * (a1 + X * (a2 + X * (a3 + X * (a4 + a5 * X))))

    ***constantan is typically 55% copper and 45% nickel

    1. And the Amiga 1000’s has a Motorola 68000 CPU which does not have a floating-point unit, so all the calculations would have been done using a software floating point maths library. I’m guessing at about 24000 FLOP’s.

        1. In the video they had one A4000 that was used for development, but if you look at the machines in the rack they look like A1000’s. I’m sure that they were all upgraded whenever the change management process and available downtime allowed.

    1. Why is that a problem? There is no need to make the resolution even lower. If your internet connection cannot handle 144p then perhaps you should contact your internet provider.

      1. Looks a bit better directly on YouTube, since it’s 4:3 and Hack-a-Day is cropping to 16:9 for some reason. Though this is clearly worse-than-VHS quality, for whatever reason. Proably a poor choice made in how it was digitized from VHS or whatever.

    2. Some historical perspective from a retired editor. Back then (in the 80s and 90s), corporate video was not even in SD it was usually still shot and edited on 3/4″ analogue tape. This video looks like the archive material may even have been from VHS!

      Why don’t you have a look here? This is in the description of the link above.

      “For a 11min version and high quality, click here…”

      1. PS There is also a nasty habit that corporate PR people have, if they don’t like your channel, production company, or director, they will send you the LOWEST quality version of the material you asked for. Even when you know there is a high quality version available, and you are prepared to pay for it.

        In my previous life I edited a documentary about a famous TV producer. It was for an ITV company, for whom he done most of his work. He had produced a famous doc for the BBC, and I knew there were broadcast quality versions available, because I had checked and got a quote for the clips we needed. So I ordered the clips, sent the bill to accounts, and what did I get? A VHS WITH BURNED IN TIMECODE! Bastards! Luckily I still had friends who could help, so I was able to get a broadcast copy of the needed bit. They had charged us for the use of the clip, AND for a 1″ dub, but sent a VHS viewing copy. I already had the producers VHS viewing copy!

        You want to complaiin about 144p? You should be so lucky!

  4. Whatever does the job. We used the 68000 on some of our intelligent terminal server boards at Data General in the 80s.
    Its architecture and instruction set reminds me of the PDP-11 and there’s some indication that was intentional.

  5. In the 90s there was a terrible war between Apple and Microsoft. As a freelance editor I needed to be able to work any system my clients had. The systems for tape editing were dominated by Sony, who were as bad as M$ and IBM for bullying buyers. The newer digital systems were dominated by Avid, from the US and running on Macs, and Lightworks, from the UK and running on PCs. You could pretty much take an Avid out of the box, plug in some external SCSI drives, turn it on and it would work. The early versions of Lightworks were a bit flaky, until the settled on an expensive motherboard and custom PSU and case, when it became just as stable as Avid. I would work with either, though I preferred the Lightworks interface, and their software people. Whenever I attended a trade show, especially if I was demonstrating, M$ reps would show up to try an persuade us to use NT workstations. Even in the late 90s Lightworks still ran on DOS and had it’s own GUI. There was a time when IBM owned the mainframe market, and there was a phrase along the lines of “the finance department things that all computers come in blue boxes with IBM on the front”. Both Sony and M$ followed that paradigm. During the last decade, however, NASA have developed there own spins of Linux to follow on from the Amigas. The Perseverance/Ingenuity mission on Mars has at least two computers running Linux for exactly the same reason they used Amigas in the past. Open Source!

  6. These are exactly the reasons why I loved the Amiga! Powerful computer that didn’t cost a lot. I remember a local computer I was invited to and local PC rep called it a game machine. This was in a world where they had bleeps and bloops for sound and CGA graphics. Of course, as soon as the PC could do the graphics and sound (which was a full blown 10 years later), they labelled it multimedia. Because it wasn’t part the PC/MAC culture, the masses paid the Amigo no attention, but for those of us who lived through it, we always knew what the best computer was. The graphics were so smooth too and you could tell when toaster graphics were used on TV. The OS was so easy to configure with Arexx to help automate common functions. I loved using SnoopDOS to analyze system function calls so I could trouble-shoot problems with software and I even ran a C-Net BBS on an Amiga for 4 years. Easily the best computer I ever used (with many kudos to the C64!)

    1. Suppose “lots” rather depended in an era were the majority were costly*, hence why Commodores, Apples, and Ataris were popular.

      *And now I’ll have to find my old copy of Computer Shopper…somewhere.

      1. I was thinking bout Computer Shopper the other day. Nearly major metro telephone book size at one point in it’s history… did ya know it’s been around since 79? I would love to see some of the earliest editions of that.

        1. Bought my first Amiga A1000 package from Computer Shopper, then bought my A2000 package from it, and all my 3.5″ floppy disks… so many floppy disks. I still buy my stuff from B&H just like I did back in the 80’s.

  7. I was surprised that the embedded rocket instrumentation had such minimal signal conditioning that the telemetry post-processing had to do high-order polynomials. Even for very tough environments (weapons systems, M1 tanks, arty, LAVs, AAVs) the lab I worked at would transmit fully scaled data, but not necessarily in engineering units such that I never used more than a 2d order polynomial (which was mostly for calibration). While I am certain that we did not have the channel count (typically 40 to 120 channels @ 10Hz to 250Hz sample rate), we did not see any decrease in data channel reliability. The big difference was that, unlike a rocket launch, we were streaming data for days or weeks, 24/7 during long-term tests at scenic and serene sites such as Yuma and 29 Palms and Irwin in the summer and Maryland and Nevada and Alaska in the winter.

    Anyone have any insight as to the design decisions of why NASA streamed raw transducer and sensor data with minimal signal conditioning?

    1. I can think of a few things:
      Less processing means less processing power i.e. less hardware less weight. More complexity = more things can go wrong. Raw data stream also means that you can apply *new* signal processing algorithms at the same data at some point later date. You can’t do that with the preprocessed data.

    2. If you do less in the spacecraft it’s lighter, uses less power, the less there is to go wrong and the more options you have with the raw data back on earth where you can use as much power and as many computers as you like.

  8. In the older spacecraft the telemetry was usually gathered by a separate subsystem which also generated most of the downlink telemetry. You needed a way to debug a spacecraft if the processor went nuts. The main computer had access to the telemetry subsystem and would then process the data internally (not downlinked unless you asked for it).

  9. I don’t understand. How would PCs released in 1995 be not fast enough. There were Pentium processors running at 200MHz with full co-processors. Compare to Amiga 4000 running at 25MHz.

    1. Commodore was always going to collapse – Once the bubble goes up for business machines to PC compatibles, all the employees were stealing software from their employers for use at home to gain a job advantage. At the education level, Apple was hammering hard at the schools. Commodore and Atari had nowhere to go. CBM went a little quicker than they might have once the C-level saw the writing on the wall and cashed themselves as much as possible, cutting all useful development funding. Such a shame.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.