Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!ut-emx!nowhere!sking
From: sking@nowhere.uucp (Steven King)
Newsgroups: sci.space.shuttle
Subject: Re: New Shuttle Computers
Message-ID: <1991Mar06.063034.12021@nowhere.uucp>
Date: 6 Mar 91 06:30:34 GMT
References: <2352@ksr.com> <1991Mar4.202334.22118@casbah.acns.nwu.edu> <1991Mar5.013344.7971@umiami.ir.miami.edu>
Organization: American Anarchist Union
Lines: 54

In article <1991Mar5.013344.7971@umiami.ir.miami.edu> jdeitch@umiami.ir.miami.edu (Jonathan Deitch) writes:
>phil@eecs.nwu.edu (William LeFebvre) writes:
>> In article <2352@ksr.com>, jfw@ksr.com (John F. Woods) writes:
>> |> The Electronic Engineering Times for 25 February 1991 has an article about
>> |> the new shuttle computers which are scheduled to fly on Discovery "next week."
>> |> The new AP101S computers use static-RAM memory and Schottky logic, replacing
>> |> the old core-memory AP101B computers.
>> 
>> Does the article say if the RAM is battery-backed or otherwise anything
>> in place to make the memory non-volatile?  One standard procedure is
>
>Static RAM memory, I believe, is non volatile memory.
>

   No, static ram simply doesnt require the refresh cycle that dynamic
 ram requires. It is still volatile -- you'll lose the contents if it
 loses power.

   While its encouraging that NASA is finally moving beyond core memory
 for the shuttle ( welcome to the 90's ), why use writeable memory for
 program storage? As their code is extremely stable, it shouldnt be any
 problem to put everything they use into masked roms without any great
 penalty in weight or size ( mask roms have much better density than
 static rams ) with it all non volatile, non writeable.

   With the 230 lbs they saved they could put a couple hundred meg disk
 array on each CPU and...  Ooops! A technology must be atleast 10 years
 old before they trust it on the shuttle...

   The sidebar to the article was atleast as interesting as the article;
 It concerned the testimony of Dr Lewis Branscom to the House Science,
 Technology and Space Committee.

      [ ... ]

       "I was startled to discover that NASA did not allow any
      hardware to fly that was not previously flown--we sold them
      1968 technology" said Branscom. "I was even more shocked
      that NASA declined to give an MTBF [mean time between
      failure] for each of the five [general purpose computers that IBM
      sold NASA]. They would not give us a spec, but instead a goal.
      I believe it was 1,640 hours. We at IBM doubled that for internal
      guidence and peace of mind." said Branscom. In contrast, he
      noted that "the mean time between failure for a $2,000 computer
      for your home carries a commercial spec of 10,000 hours."

      [ ... ]

   ( and is a lot more powerfull as well )


-- 
Look Ma! No .sig!
                                        ..!cs.utexas.edu!ut-emx!nowhere!sking