11 Feb
2015
11 Feb
'15
10:51 a.m.
On 02/11/2015 12:38 AM, Chuck Guzis wrote:
On 02/10/2015 09:56 PM, Jon Elson wrote:
I'm sure not aware of that. #2 ESS uses ferreed switching
elements, but it was WAY after
core memory was developed. The original core design was
done somewhere around
1948 - 1950, I can't find an exact reference. I'm unaware
of any similar, entirely
electrical/magnetic latching scheme used in telephony that
predates that. Western
Electric crossbar was quite mechanical although it did use a
matrix scheme to
select crossbar contacts to opened or closed.
Oh, absolutely! There was a lot of work on using
ferrite
rings as storage
and logic elements at that time, but Forrester and Papian
really
extended what had been done before, and the coincident
current
scheme was really ELEGANT and made large arrays of fast
memory
practical. The bigger you built the array, the more
memory you got
with small increments in the number of drivers.
Didn't coincident-current relays come before that (as
used, for example, in telephone switching equipment)? So
the basic idea was there.
I've always been fascinated by magnetic core
logic; both
using "hard" magnetics (e.g. Univac SS) and "soft" (e.g.
Parametrons). I wonder if magnetic core for memory hadn't
been developed, would we have developed electrostatic or
some other technology to the same density?
Would we have developed ultra-fast recirculating memory?
I'm sure memory technology could have gone a number of
different ways. i don't
think classic core memory could NOT have been developed, it
is too marvelous
for people to have missed the concept for much longer.
There were some other schemes, twistor and plated wire
memory, and biax memory.
Some of these were non-volatile, which was an advantage to
the military. But,
one of those technologies could have become dominant, instead.
I never understood why bubble memory didn't continue to
progress. Vertical
Bloch line memory might have eventually developed to the
capacity of
modern flash memories, and probably not had the wear-out
problem.
They never would have reached the read performance of flash,
but might
have kept up with the write performance.
Hmmm, flash memory IS really electrostatic. And, a lot
better than anything
like Williams tubes. I guess old-style DRAM was basically a
silicon implementation
of a Williams tube!
Jon