NASA Not Sure If Space Shuttle Computers Understand New Year's
from the can't-figure-this-out-yet? dept
It’s no secret that the space shuttle is now based on very old technology. In fact, even NASA admits that it’s the equivalent of an old pickup truck. In fact, at times, NASA has turned to the likes of eBay to pick up old pieces like 8086 chips to replace original parts on the shuttle. So, perhaps it shouldn’t come as a huge surprise to then find out that NASA isn’t at all comfortable that the shuttle’s computers can survive New Year’s in space. Yes, that’s right. The folks at NASA have worked hard to make sure that a space shuttle is never in space from December 31st to January 1st of any year, for fear that its computers would go haywire in a Y2K manner — perhaps causing serious damage to the ability of the shuttle to continue its mission. Still, you would think they’d be able to, you know, test that sort of thing out — but NASA says they simply have no idea what would happen, and they’d like to avoid finding out. They’re not all that worried, but a statement like the following hardly seems confident: “if we have an ‘Oh my god,’ and we have to be up there, I am sure we would figure out a way to operate the vehicle safely…. It just wouldn’t be flying in the normal certified mode that we are used to flying.” If I were one of the astronauts, hearing “I am sure we would figure out a way…” can’t be the most comforting of thoughts.
Comments on “NASA Not Sure If Space Shuttle Computers Understand New Year's”
Just fix it
Perhaps they could rewrite the code and fix the bug?
It’s not beyond a 12 year old is it?
I thought these NASA types considered themselves smart people.
(before offering some convoluted obfuscatory argument about how amazingly complicated and difficult it is to fix this bug please be aware you are just digging a deeper hole and making NASA look even more stupid for adopting an unmanagable complex software solution where a simple one would have done the trick)
Re: Just fix it
not sure if you were around when the space shuttles were originally built, but computers back then weren’t exactly the most fun thing to program for, especially integrated systems where the only option was direct assembly. There was no .net that does everything for you.
More to the point, legacy systems are a bugger to maintain, even if built brilliantly. And although I’m sure nothing would happen come new years, would you really want to risk it with a several billion dollar machine and several human beings?
Re: Just fix it
The shuttle was designed in the 60’s and build in the 70’s. The thing NASA wanted to avoid was to put the cost of testing on 4 shuttles at the time, but now it’s 3 but it’s still costly.
correction
The above quoted statement did not come from NASA. It was Boeing that made the statement.
http://www.cnn.com/2006/TECH/space/11/09/space.shuttle.ap/index.html
Is it that hard to replace the hardware inside a space shuttle. And if it is, isn’t it at least the upgrade ‘worth it’. Running a platform that is a lot more flexible and stable, on a hostile environment like space, is worth the $$ to me.
On the other hand, they might be afraid of upgrading because of BSODs, I’im sure they don’t get those in their antiques 😛
I understand this only too well (unfortunately)
About the time the first space shuttle was launched I was building my own computers from 6800 microprocessors and writing my own operating systems for them in assembly 😉
It’s important to understand that ALL systems are destined to become “legacy”. That is the point of the software lifecycle model which you can understand by reading Sommerville. Quality software, of the kind you would expect to find in a space shuttle, takes this into account from day zero. Military and aviation code is invariably specified in Z, written in ADA and passed through numerous program proving tools. Concepts like .net are not even in the same ballpark of computer science.
It’s most likely a hardware problem involving width limitations of 16 bit floating point numbers. JFYI, to my knowledge the General Dynamics Tomahawk (cruise missiles) that carry nuclear payloads also suffer the legacy limitations of the Motorola 68000 hardware, but you don’t hear many people spreading fear stories about those mixing up their timezones on Jan 1st and taking out New York instead of Iran.
Given that and my own experience of such systems I would strongly stand by two statements,
1) the existence of the bug is 1% technical and 99% human negligence/ economic shortcuts.
2) Resolving it is trivial (given the available resources and expertise) and it should have been quietly fixed instead of becoming a “political” story to embarrass NASA.
Dmitriy, it’s irrelevant who made the mistake, the mistake was made.
Re: I understand this only too well (unfortunately
Really? They have no budget, do you just not launch on a suspected date, or do you spend precious money running down a bug?
Which one costs less?
Looks like we need some space cowboys.
floating point, schmoting point
Hey folks, I programmed military systems back in 1981 and I can tell you that we didn’t use the floating point capabilities of those old chips- too slow.
Instead, almost everything was done with fixed point- sometimes double precision fixed point for complex calculations- and programmers kept track of where the decimal point would be for the output based on where it was for the inputs. (Hope no old Soviets are reading this… Nevermind, it wasn’t classified even back then. :)) )
Re: floating point, schmoting point
I guess you never programmed the 1750 16-bit ISA processor- it was blazing at floating point. 🙂 Though I did do a lot of fixed point scaled calculations like you described!
Open the pod bay doors hal...
..please? Come on, open ’em up…..
I’ll donate my Pentium 133 to upgrade their 8086 if it will help the space program.
Why fix it?
This just proves futher that the shuttles now only have one place in aviation: In a museum. The shuttles were not designed to be in service for as long as they have. Look at it this way, if you were to compare the shuttle to a car, you know that after roughly 30 years, things are going to be problematic, and if that “commute” costs billions of dollars and the lives of numerous astronauts, you wouldn’t leave it to that old Chevelle. Time to put them in a museum and get some new hardware. Maybe we should look to Toyota to make the next space shuttle….
I don’t care what they do, also long as Microsoft isn’t writing the programs of any space shuttle. Cause that would bring a whole new meaning to blue screen of death.
I understand this only too well (unfortunately
wow, somebody sounds like they just spent a year in a computer science course… or are you on your 3rd year?
Why the 8086
I remember in high school, lo these many years ago, hering that the shuttle uses the 8086. (this was when the 80486dx was just about to be released) Many of us wondered why. The reason, for the most part, as to why it has not been upgraded, has nothing to do with know how, but more to do with unknowns.
As the technology has gotten better newer and faster, it has also gotten smaller. It now takes less energy to switch a tranistor from off to on than it used to. A pentium chip uses less energy per transistor than does the 8086. The truth is that NASA has no idea what might happen to the onboard computers if a solar bust or another “Spave thing” (my term) came through. If it powers up enough trasistors, or shuts them off, it could mean disaster. Using the High energy chips makes sense.
There is really no energy sheilding on the shuttle, to do that would make it too heavy. Rather than spend the millions of dollars it would take to figure out what they could upgrade it to, they have opted to stick with what they know works.
If I had to guess, the new (ugly, POS) rocket they are coming up with will use something a little more robust.
Time for an upgrade
According to what the shuttle computer specs say, you could run the entire thing off a cheap laptop. I say go buy a cheap laptop.
Any of you who say “upgrade it, jeeze.. even i can think of that” really dont understand the situation.
The shuttle, similar to say, a pentium 133, cannot ‘just be upgraded with the newest hardware…
You cant throw an amd x2 into a pentium 133 , and you cant just throw a newer chip into the space shuttle…
And for those of you who dont understand the Y2K problem, the problem is that to save space (back when computer memory was measured in bytes… maybe KB) the year was stored as a two bit number, instead of 4, as that made sense in the day, no one thought 30 years in the future when 99 became 00.
Luckily by the time Y2K rolled around, computers had GB’s of memory, so there was no problem to re-code everything to use 4 bits of memory.
Same with the shuttle, i doubt that the shuttle’s main systems have GB’s of memory, as it wasnt around when it was desinged. To save space, they dont store the year at all, just the day count, 1-365.
They are unsure the effects of going from 365 -> 1, and dont really care to find out…
So this problem isnt easily fixed with more memory, its just a scenario that hasnt been tested.
As far as resolving it, theres no way to really know the effects without testing it… You can read code a thousand times, and still miss something that your not looking for…
NASA Buyers - "SALE"
Hey Nasa guys!!
I’m selling some old stuff that be very usefull for you!!
4 brand new old 486 DX2 66
2 Pentium 100Mhz
And… tchan tchan tchan tchan !!!!!
1 PC AT 8086 – With 8087 co-processor !!!! Thats’s amaizingly fast for your navigational calculus!!!
And, YES we have too, some great matricial printers !!
Need a Green Phosp. CRT monitor… You came to the right place !!!!
Just a joke… BTW, if the text above is true, they deserve!!!!
Hehehehehehe!!!!
Cheers~
To be crystal clear
Important to say,
I love NASA and all that NASA does and represents.
But sometimes NASA adm do thing that is very hard to understand.
Despite all theories about how hard or ease is to upgrade a PC based computer the core point is about “how the guys there don’t handle with a technology that double the capacity every N months (More).
The Energy consuption is not a key role either, since you’ll use less processor to do the same work faster if you change of an old 486 to a new AMD/Intel about 3GHZ.
Of course the Operational System and the Systema would be renewed too. But. c’mon these is the minimum. Every company are doing it al the time. Is part of the game.
PS. Old stuff is more expansive than new one.
This isn’t a bug — I suspect they know that it doesn’t work because of the fact that it was specifically made “not a requirement” back in the day.
I suspect it’s not the fix that’s the hard part — its the validation and testing of the software that will take more than the 2 months allowed.
The shuttle isn’t something that you slap code in and go fly. It’s software has received accolades as the most bug-free software ever written, especially considering that it was originally written back in the 70s — and extended ever since.
It’s a legacy code base with a necessarily long path to validate and verify the code for a flight.
As for those guys who say “just upgrade” — dudes, the environmental and vibration environment for all the equipment in the shuttle is ridiculous.
I am glad NASA is conservative here — I mean, it’s great. The software guys go “I think we shouldn’t fly with this bug”, and people listen to them! When does that happen in real life? It’s usually “ship now, patch later”.
A simple Solution?
Why not set the clock on the shuttle 6 months
ahead, when flying during December. This way the
risk of an ‘end of year’ problem disappears.
It should not really matter to the other operations.
If this is a problem set all the clocks on the ground also
ahead, to the same time as on the shuttle The system should work fine.
They can save the money with this solution to get new computers.
right
“Should” work fine is not good enough, folks. It has to work. Do you want to stake the lives of 7 people and a billion or so dollars on “should”?
It'll be just fine
I happened to be standing out watching the Challenger when it lifted off on a freakishly cold winter day when the engineers said don’t launch and the people who didn’t know the system said it’ll be just fine. And it was–until that damn O-ring broke–but that wasn’t supposed to happen. Yes, the shuttle should have been replaced years ago, when it was intended to be replaced. But the money to replace it wasn’t there [thank you, Carl Sagan!]. Then when they throw a little money at NASA they want immediate results.
John Kennedy called for putting a man on the Moon by the end of the ’60s–and they made it with less than 6 months to spare, with the whole country behind them. The government and the country isn’t giving NASA time to upgrade or replace anything, so they have to go with what they have. They’re more interested in keeping people alive than going with the latest technology when they aren’t given time to verify it.
We’ve lost so much time in our attempt to get into space–but now is not the time to try to make it up by doing things that take a chance with people’s lives.
Six Months
the six month idea seems fine, since the shuttle would not be conencted to anything except NASA computers, which wou;d ba a small limited network, so changing all the dates would not cause any obvoius problems. After all, who says this is november? It would be no diffrent to them simply setting all the dates wrong when they set up the system.
Old Technology
Many moons ago, I worked for Burroughs that made a lot of government spec hardware. The chips used were Military spec, and much more resilient than commercial grade chips. In fact, I remember working on a machine that some Bozo had put a chip in backwards, thereby reversing the supply voltage. The chip was very hot. Hot enough to burn your fingers. I didn’t have a spare, so I tried turning it back around, and it worked perfectly. As technology developed, it became prone to static damage. Shortcuts in the manufacturing of floating point processors left a ‘bit bobble’ that had to be manually corrected, or allowed for. In something as precise as space travel, I don’t think that ‘bit bobble’ could be tolerated. Motorola wasn’t the only one to repeat errors in their chips. Intel had a flaw in the early chips (286?) which was corrected. It’s my understanding that they used the original flawed masks to make a multiple chip on a chip (386, 486, Pentium) replicating the flaw many times over. If you had a design that you spent a ton of time and effort debugging and perfecting, would you want to chart new waters with “newer technology?”
Time and the 8086
Oh, yeah. As far as I remember, there is no reference to time in the processor’s machine code. That is handled by the operating system. These processors were used in clusters of 5 and each one checked the others’ operation. I really don’t see how Y2K would come into play. I’m sure that the Shuttle wouldn’t just say “it’s 1907, I haven’t been invented yet.” Are you sure they don’t fly in December because of the weather, or end of the year budgets?