ISS computers only recently upgraded to Pentium processors

Iwannahurl

Astronaut Dan Tani, who's in Ireland at present, was interviewed by Pat Kenny this morning.

He mentioned that the International Space Station's computers were only recently upgraded from 386 to Pentium processors.

I'm aware that, historically, successful space missions have been conducted with remarkably little computing power.

http://www.ibiblio.org/apollo/index.html

Still, I found it surprising that the ISS was depending until recently on computer technology from the 1980s.

Have they leaped from the 1980s to the 21st Century? Or have they settled for a decade-sized jump to 1990s-level Pentiums?!

Will the upgrade make much difference?

AugustusMinimus

Specalised software used may not have been compatible with later intel chipsets.

thecommander

Most of the systems have be tested and tested again. Switching to new hardware means that they'd have to retest all the software up there, and then test to make sure that it's compatible with older hardware. Not the best place to have a blue screen of death.

Zubeneschamali

The emphasis in NASAs spaceships is reliablity. They don't want the latest/greatest computer technology, they want something that absolutely positively will not crash. If it's programmed in an 80s language using 80s tools, good. If the software was actually written in the 80s and has 30 years of testing behind it, even better.

Popoutman

Radiation hardening is much harder to do (well|cost-effectively) on the smaller process cpus.

tricky D

There's also the long development time. The ISS was started in 93, about the same time Pentiums were released. As already noted, this kind of development avoids early adoption due to reliability factors.

Also notable about the 386 was the separate maths coprocessor needed for some calculation-heavy systems eg. AutoCAD.

Seanie M

Popoutman wrote: »

Radiation hardening is much harder to do (well|cost-effectively) on the smaller process cpus.

This is the big reason why chipsets take so long to be updated. It was only in the late 90's that the shuttle fleet was upgraded from its array of 8086 processors to the then-equivalent of Pentium III's. The chips are bulky due to the radiation proof casings on them, and once the entire 'dashboard' is/was in place, it's not supposed to be tinkered with for some time.

Even though the ISS began assembly in 1998, the design started in 1982 (Japan and America) and main construction began in 1993, having begun using parts that began their construction in the 80's. Don't forget, the initial completion date, back in the 80's was for 1999. It was only last year (2011) that the final module was added.

Seanie.

Alun

I worked at ESOC in the 1980's on a couple of scientific missions (EXOSAT and GIOTTO) and in both cases the processor power employed was truly primitive, even by the standards of the day. The main DHSS (Data Handling Subsystem) and AOCMS (Attitude and Orbit Control and Measurement System) systems in GIOTTO, for example, were both powered by space hardened versions (built using a special silicon on sapphire process) of the RCA 1802 8-bit microprocessor, first introduced in 1976! They even underclocked them by a considerable degree to enhance reliability. I actually worked on the GIOTTO real-time simulator, and we simulated both subsystems using a software 1802 emulator written in FORTRAN running the actual spacecraft code, it was that slow.

Some of the on board experiments provided by third parties were a little more adventurous though, I seem to remember one of the experiments on EXOSAT was powered by a Motorola 68000, for example, which was way more powerful than anything on the spacecraft itself, but for the spacecraft itself, reliability was paramount.