Iggy/Chris -
The STS (Space Shuttle) actually uses a IBM Federal Systems-build redundant
computer system called the Advanced System/4Pi Model AP-101 computer around
1978-1979. (And yup, there was a 3/Pi.) Not a 1750x clone, but was 32-bit
rad-hardened system, and expensive. Some of the work was done here in
Boulder Colorado and some IBM overview stuff can be found at
http://www.ibm.com/IBM/history/timeline.nsf/products3. Actually a pretty
good system, with I believe four riding in the original STS design using
voting as main redundancy check. Several other interesting IBM "gummint"
projects have taken place in Boulder, but...
Bruce
bkr_at_WildHareComputers.com
www.WildHareComputers.com
www.SimuLogics.com [Novas are forever...]
----- Original Message -----
From: "Chris Kennedy" <chris_at_mainecoon.com>
To: <classiccmp_at_classiccmp.org>
Sent: Saturday, February 03, 2001 11:14 AM
Subject: Re: Fwd: Naked Mini Broke
> Iggy Drougge wrote:
>
> [snip]
>
> > What's wrong with old-arsed computer anyway???
>
> Particularly in space-based applications, where flight qualification for
> man-rated hardware is a pangalactic bitch.
>
> > BTW, I read somewhere that the space shuttle actually used core memory,
> > apparently due to the bad resistance towards radiation in older IC
memory.
>
> I'm sure that's a component -- even contemporary dynamic memory is less
> rad hard than you'd probably like, but that's hardly the overriding
reason.
> My increasingly dim memory suggests that the processor are slightly
modified
> 1750A machines. There was some indication that they were a derivative of
the
> processor using in the F-16, but I never was able to verify that. Being
'70s
> designs suggests that core would have been the only real option.
>
> > Could anyone confirm whether this still holds true?
>
> I have no idea if the processors are being swapped out as part of the
> flight deck modernization program. If so it can't be too radical a
> departure from the existing architecture, since only one orbiter has
> completed that modification cycle and the thought of having two
> sets of tools for building the code loads for the orbiters staggers the
> mind.
>
> Oh yeah, that's the other thing. The code load for each orbiter is unique
> to the mission, and there's no real executive function -- it's a bunch
> of cooperating real-time tasks, such that if one pukes out it probably
> takes out the rest. There's been at least one documented case where
> the flight code load, during on-the-pad prelaunch simulation, dropped
> the ball and the flight deck displays ended up with the shuttle's
> answer to the BSOD -- an 'X' drawn through the flight deck displays.
>
> If you think that's odd, you should check out the way that voting works...
>
> --
> Chris Kennedy
> chris_at_mainecoon.com
> http://www.mainecoon.com
> PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97
Received on Tue Feb 13 2001 - 20:55:21 GMT