Path: utzoo!attcan!uunet!husc6!bloom-beacon!bu-cs!purdue!decwrl!labrea!glacier!jbn
From: jbn@glacier.STANFORD.EDU (John B. Nagle)
Newsgroups: comp.misc
Subject: Re: Core memory
Message-ID: <17561@glacier.STANFORD.EDU>
Date: 18 Jul 88 17:11:36 GMT
References: <1486@crete.cs.glasgow.ac.uk> <1010@garth.UUCP>
Reply-To: jbn@glacier.UUCP (John B. Nagle)
Organization: Stanford University
Lines: 16


      Most core memory required three wires through every core: X drive,
Y drive, and sense.  The X and Y lines are straight, and were usually put
in place with semiautomated techniques similar to weaving.  The sense wire,
which zigzags through every core in an array, usually had to be threaded though
the cores by women using tweezers, looking through microscopes.  UNIVAC
made core memory in this way through the mid 1970s, in Roseville, MN.
Memory prices for that technology ran about $1M/megabyte.

     IBM pioneered automatic core-stringing machinery in the 1960s, but
for a long time, the cores had to be bigger for automatic stringing, which
made the entire memory system much larger.

     I could say more, but it's a dead technology.

					John Nagle