Well, you know what they say about age and memory ... ;)
Actually, this goes a bit beyond what I actually forgot, lol:

---------- Forwarded message ----------
From:  Tony B. Shepherd <spamless>
Date: Jul 21, 2007 2:03 PM
Subject: Re: [HP3000-L] JCL
To: [log in to unmask]

Pete Eggers wrote:
> Ah, the 64k memory box...  Memory was called "core" not "RAM" back then.
> Why?  Because every memory bit was a tiny magnetic donut called a "core",
> hand strung at the factory with 2 wires -- one horizontal and one vertical.
> The refrigerator sized cabinet not only housed this core memory, but a
> large oil tank and pump to keep the core memory cool.
{ Rest snipped }

Hi Pete - dedicated lurker here :)

Just wanted to say your memory is close, but not quite :)

There were 3 wires.

 Two wires were in the X and Y planes of memory cores, and the third wandered
around a bit (pun intended) depending on the design. Take a look at any of
the old museum close-ups, and you'll see diagonal wires in the mesh.

I don't remember the exact values for voltage and current for IBM's 360/30
and 360/40 (where I worked), but it went something like this:

It took an 8 volt pulse to establish the polarity of a bit. Assuming a 100
bit core (10 by 10), 5 volts was applied to one of the ten X lines, and 5
volts was applied to one of the ten Y lines. Eighteen bits (nine vertical
and nine horizontal) saw 5 volts, but only the intersection bit (tenth in
each axis) saw 10 volts, and would be set.

Reading a bit was called "destructive readout". Writing a core bit to zero
or one was controlled by the polarity of the X and Y lines. To read a bit,
the hardware tried to set it to zero. If it was already zero, the sense
(third) wire wouldn't see anything. If it was one, changing it to zero
induced a pulse in the sense line, and an additional memory cycle was needed
(to make it a one again).

Since you never knew at power up the memory state, all power up procedures
includes setting memory to known values ("clearing core" was one term). Many
machines did that automatically, but the IBM 1620 had a short (16 digit?)
sequence typed in on the console to clear memory.

Those were indeed interesting times :) We've come a long, long way but in
many ways nothing has changed.

Hope I helped the recollection gently - thanks for the trip down memory
lane. If you quote to the list, please omit my eddress - my spam levels are
just fine as they are :)
--
Regards  --  Tony B. Shepherd  --



Thanks Tony!

The depth of knowledge, especially scientific knowledge, on this board
never ceases to amaze me!  I've been spending too much time with
<affectionately ;)> GUI Guppies!

- Pete

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *