Blast From The Past - Core Rope Memory

[B1ackwater]

Computer systems for govt/military projects tend to
be contracted for the best tech - for when the contracts
were drawn up. This can be YEARS before the actual
products are delivered.

Case in point, the Apollo Guidance Computer ... you
know, the thing Tom Hanks was laboring over just before
they bailed to the LEM ........

I get the impression the contracts were drawn up in '49.

Magnetic core memory, a matrix of tiny magnetizable doughnuts,
was usually produced to be in a meter-square box with a mass
of crossing wires with a core at each crossing. A cubic meter
might get you a few K of memory.

"Rope" memory was HAND WIRED, a string of wires with those
little cores carefully positioned all along the length. Wiki says it
was sometimes called LOL Memory - "Little Old Ladies" - for the
women who meticulously constructed the assemblies.

This approach offered nearly 10 times the data density per
cubic-foot than the commercial boxes. The downside was
that you'd better damned well get it right the first time, you
don't edit a rope memory, you reconstruct it from scratch.

Now clearly, for a space-ship, 10X the data density per cubic
whatever WAS very attractive - and for NASA the costs were
pretty irrelevant.

Thing is, by '69 the notion of pluggable fuse-programmable PROMs
was far from alien. Higher data density plus pretty much total
resistance to magnetic/radiation corruption.

However the CONTRACT was from (many) years past. So, we
got rope, not PROM chips.

For newbies, a "fuse-programmable PROM" is a write-once
memory chip that is basically a collection of fine wires. To
write the thing you apply enough current to actually burn away
some of those tiny wires, leaving the matrix full of '1''s and '0's.
These were later replaced by other kinds of permanent memory,
UV-erasable EPROMS and such. (still have a short-UV box to
erase those btw :-)

You can STILL buy fusable-link PROMS from a few sources
however .... if it HAS to be armor-plate tough ........

In the mid 80s I got a tour of an attack submarine. The interesting
part was the sonar cubicle. In there was a discrete-transistor
computer about a cubic meter (think IBM-360) and a few of
those "hard drives" where you had ten or so big disk platters
in a plastic case. You could put the cover on and remove the
whole disk assembly (but make SURE the thing had actually
spun down !). The data density was fer-shit. The speed was,
well, a speck better than spools of mag-tape. This was how
we were to defeat the Russkies/Chinese in the 80s.

Again, we had a fairly new govt project, but the hardware had
been contracted many years before and was already totally
obsolete by the time the final product was delivered.

[Tauno Voipio]

Post by B1ackwater
Computer systems for govt/military projects tend to
be contracted for the best tech - for when the contracts
were drawn up. This can be YEARS before the actual
products are delivered.
Case in point, the Apollo Guidance Computer ... you
know, the thing Tom Hanks was laboring over just before
they bailed to the LEM ........
I get the impression the contracts were drawn up in '49.
Magnetic core memory, a matrix of tiny magnetizable doughnuts,
was usually produced to be in a meter-square box with a mass
of crossing wires with a core at each crossing. A cubic meter
might get you a few K of memory.
"Rope" memory was HAND WIRED, a string of wires with those
little cores carefully positioned all along the length. Wiki says it
was sometimes called LOL Memory - "Little Old Ladies" - for the
women who meticulously constructed the assemblies.
This approach offered nearly 10 times the data density per
cubic-foot than the commercial boxes. The downside was
that you'd better damned well get it right the first time, you
don't edit a rope memory, you reconstruct it from scratch.
Now clearly, for a space-ship, 10X the data density per cubic
whatever WAS very attractive - and for NASA the costs were
pretty irrelevant.
Thing is, by '69 the notion of pluggable fuse-programmable PROMs
was far from alien. Higher data density plus pretty much total
resistance to magnetic/radiation corruption.
However the CONTRACT was from (many) years past. So, we
got rope, not PROM chips.
For newbies, a "fuse-programmable PROM" is a write-once
memory chip that is basically a collection of fine wires. To
write the thing you apply enough current to actually burn away
some of those tiny wires, leaving the matrix full of '1''s and '0's.
These were later replaced by other kinds of permanent memory,
UV-erasable EPROMS and such. (still have a short-UV box to
erase those btw :-)
You can STILL buy fusable-link PROMS from a few sources
however .... if it HAS to be armor-plate tough ........
In the mid 80s I got a tour of an attack submarine. The interesting
part was the sonar cubicle. In there was a discrete-transistor
computer about a cubic meter (think IBM-360) and a few of
those "hard drives" where you had ten or so big disk platters
in a plastic case. You could put the cover on and remove the
whole disk assembly (but make SURE the thing had actually
spun down !). The data density was fer-shit. The speed was,
well, a speck better than spools of mag-tape. This was how
we were to defeat the Russkies/Chinese in the 80s.
Again, we had a fairly new govt project, but the hardware had
been contracted many years before and was already totally
obsolete by the time the final product was delivered.

There is a difference between normal core memory and a rope memory:
the normal core memory uses permanent-magnet material and the rope
memory uses transformer cores with minmal hysteresis.

A rope memory is an early ROM. If a word line goes through the hole,
it is a '1' and if it goes around the core, it is a '0'. The rope
memory cores cannot be x-y matrixed in a similar way as the normal
high-hysteresis memory cores.

--

-TV

Been there, done that, a 24-bit wide boot loader rope memory in the
late 1960's.

[B1ackwater]

On Wed, 6 Jan 2021 13:20:33 +0200, Tauno Voipio

Post by Tauno Voipio

Post by B1ackwater
Computer systems for govt/military projects tend to
be contracted for the best tech - for when the contracts
were drawn up. This can be YEARS before the actual
products are delivered.
Case in point, the Apollo Guidance Computer ... you
know, the thing Tom Hanks was laboring over just before
they bailed to the LEM ........
I get the impression the contracts were drawn up in '49.
Magnetic core memory, a matrix of tiny magnetizable doughnuts,
was usually produced to be in a meter-square box with a mass
of crossing wires with a core at each crossing. A cubic meter
might get you a few K of memory.
"Rope" memory was HAND WIRED, a string of wires with those
little cores carefully positioned all along the length. Wiki says it
was sometimes called LOL Memory - "Little Old Ladies" - for the
women who meticulously constructed the assemblies.
This approach offered nearly 10 times the data density per
cubic-foot than the commercial boxes. The downside was
that you'd better damned well get it right the first time, you
don't edit a rope memory, you reconstruct it from scratch.
Now clearly, for a space-ship, 10X the data density per cubic
whatever WAS very attractive - and for NASA the costs were
pretty irrelevant.
Thing is, by '69 the notion of pluggable fuse-programmable PROMs
was far from alien. Higher data density plus pretty much total
resistance to magnetic/radiation corruption.
However the CONTRACT was from (many) years past. So, we
got rope, not PROM chips.
For newbies, a "fuse-programmable PROM" is a write-once
memory chip that is basically a collection of fine wires. To
write the thing you apply enough current to actually burn away
some of those tiny wires, leaving the matrix full of '1''s and '0's.
These were later replaced by other kinds of permanent memory,
UV-erasable EPROMS and such. (still have a short-UV box to
erase those btw :-)
You can STILL buy fusable-link PROMS from a few sources
however .... if it HAS to be armor-plate tough ........
In the mid 80s I got a tour of an attack submarine. The interesting
part was the sonar cubicle. In there was a discrete-transistor
computer about a cubic meter (think IBM-360) and a few of
those "hard drives" where you had ten or so big disk platters
in a plastic case. You could put the cover on and remove the
whole disk assembly (but make SURE the thing had actually
spun down !). The data density was fer-shit. The speed was,
well, a speck better than spools of mag-tape. This was how
we were to defeat the Russkies/Chinese in the 80s.
Again, we had a fairly new govt project, but the hardware had
been contracted many years before and was already totally
obsolete by the time the final product was delivered.

the normal core memory uses permanent-magnet material and the rope
memory uses transformer cores with minmal hysteresis.
A rope memory is an early ROM. If a word line goes through the hole,
it is a '1' and if it goes around the core, it is a '0'. The rope
memory cores cannot be x-y matrixed in a similar way as the normal
high-hysteresis memory cores.

Designing/wiring such a thing is beyond me - especially when
we're talking significant amounts of memory. It's amazing how
they used to do so much with so little. Very literally got us to
the moon, humans 300ky dream.

Mercury-delay line memory is easy by comparison.

NEVER diss those old white-shirt/black-tie people from
the 50s. Amazing skills, especially considering how new
the tech was at the time.

[ray carter]

Computer systems for govt/military projects tend to be contracted for
the best tech - for when the contracts were drawn up. This can be YEARS
before the actual products are delivered.

If done properly, that need not be the case. I was a Civil Servant and
White Sands (Army) Missile Range when we required upgrade from PDP-11's
for several specific applications. In the early nineties (if I recall the
time frame correctly) we installed the first DecStations in the entire
sales region.

[B1ackwater]

Post by ray carter

Computer systems for govt/military projects tend to be contracted for
the best tech - for when the contracts were drawn up. This can be YEARS
before the actual products are delivered.

If done properly, that need not be the case. I was a Civil Servant and
White Sands (Army) Missile Range when we required upgrade from PDP-11's
for several specific applications. In the early nineties (if I recall the
time frame correctly) we installed the first DecStations in the entire
sales region.

White Sands would have been a priority case - far fewer
delays.

But why would you want to replace PDP-11's ? :-)

I have a PDP-11 emulator and images of Unix to run
it .... just haven't gone through the steps yet. These
days, the emulated version will be faster than the
original.

One of my first real jobs was porting-over an LSI-11
system's data to a 1st-gen IBM-PC. Somewhere I still
have photos of the horrible wiring nightmare required
to "print" the data from RS-232 to RS-232. No, they
were not 100% compatible despite the "standard"
implied :-)

The LSI-11 had the wonderful old 8" floppies. Somehow
I doubt you can find anybody/anything to cope with those
anymore.

I saw the Smithsonian/National Archives is freaking out
about the volatility of old computer hardware & data formats.
Vast amounts of historically valuable info risks being lost
because there is no more hardware/software to cope with
all the old stuff. You can find a physical copy of the Declaration
Of Independence ... but can you read the data from the Gemini
space program ???

[ray carter]

Post by ray carter

Computer systems for govt/military projects tend to be contracted for
the best tech - for when the contracts were drawn up. This can be
YEARS before the actual products are delivered.

If done properly, that need not be the case. I was a Civil Servant and
White Sands (Army) Missile Range when we required upgrade from PDP-11's
for several specific applications. In the early nineties (if I recall
the time frame correctly) we installed the first DecStations in the
entire sales region.

White Sands would have been a priority case - far fewer delays.
But why would you want to replace PDP-11's ? :-)
I have a PDP-11 emulator and images of Unix to run it .... just
haven't gone through the steps yet. These days, the emulated version
will be faster than the original.

The new DECstations were a huge upgrade. RISC computer running DIGITAL
unix and X-windows. Much more memory and way more speed. We were running
software that analyzed radar data to determine target motion (not where
the tracked object went so much as what it did getting there - spin rate,
nutation, precession) and needed all the processing power we could get.
The DECstations were eventually replaced by DEC Alphas and later dual
Xenon RedHat systems. Each step was orders of magnitude faster than the
previous one.

[B1ackwater]

Post by ray carter

Post by ray carter

Computer systems for govt/military projects tend to be contracted for
the best tech - for when the contracts were drawn up. This can be
YEARS before the actual products are delivered.

If done properly, that need not be the case. I was a Civil Servant and
White Sands (Army) Missile Range when we required upgrade from PDP-11's
for several specific applications. In the early nineties (if I recall
the time frame correctly) we installed the first DecStations in the
entire sales region.

White Sands would have been a priority case - far fewer delays.
But why would you want to replace PDP-11's ? :-)
I have a PDP-11 emulator and images of Unix to run it .... just
haven't gone through the steps yet. These days, the emulated version
will be faster than the original.

The new DECstations were a huge upgrade. RISC computer running DIGITAL
unix and X-windows. Much more memory and way more speed. We were running
software that analyzed radar data to determine target motion (not where
the tracked object went so much as what it did getting there - spin rate,
nutation, precession) and needed all the processing power we could get.
The DECstations were eventually replaced by DEC Alphas and later dual
Xenon RedHat systems. Each step was orders of magnitude faster than the
previous one.

I still have a 3-inch-thick book on everything you need to know
about using/programming DEC-VMS. A really good system, even
now. They thought of pretty much everything. A bit pre-internet,
but still inherently set up for multi-user/processor and distributed
systems. A Raspberry Pi these days might be faster, although
it is not going to support hundreds of connected terminals
and devices scattered across the globe as well. The DECs
came with lots of semi-intelligent firmware boxes that relieved
the CPU burdens.

Gotta FIND one of those micro-VAXes... e-bay surely has some
that run.

PDP/LSI-11's can also be had, though the price tends to be steep.
The only LSI-11 I have would need a LOT of electronic components
replaced (mostly paper caps) before I'd dare power it up The 8"
floppies are another challenge in and of themselves.

My current interests though lean towards finding a running "Hot
Coco" with OS-9 installed ....

Anyway, "faster" has its advantages - but "elegant"
has its charms :-)

[B1ackwater]

On Thu, 7 Jan 2021 04:39:51 -0800 (PST), Pedro Valdez

Post by B1ackwater
I saw the Smithsonian/National Archives is freaking out
about the volatility of old computer hardware & data formats.
Vast amounts of historically valuable info risks being lost
because there is no more hardware/software to cope with
all the old stuff. You can find a physical copy of the Declaration
Of Independence ... but can you read the data from the Gemini
space program ???

I dunno about Gemini, but I remember when a local academic found some copies of Apollo magtapes,
they were 7-track. A data transcription contractor happened to have a tape drive, but it wasn't working.

That's a serious issue. While you can probably come across
one of those tape-drives, the electronic components within may
have an expiration date (paper capacitors are nefarious). The
moment you power them up the magic smoke escapes and
there's a chain-reaction .....

Then there are the hundreds of proprietary, oft forgotten, formats
for the data. The practice during the era of tiny tiny memory and
tiny tiny storage was to bit-pack ... kinda like 'c' structs where
three bits of a byte are one thing and the next ten are something
else and the next two are ..... and .... done that myself on some
old datalogger projects. There are even weirder, tighter, ways
to pack data on a limited medium and speed-vs-quantity can be
a factor too. Oh, and what do those bits MEAN ? Might not be
anybody alive who knows.

Sure, the NSA could probably figure it out, but they are too busy
with spying on everybody.

Finally there's the data media itself. Paper tape degrades, punch
cards get frayed, the coating on mag-tape and floppies flakes off.
Some old tapes simply cannot be read conventionally - the coating
will fly off as dust the moment you unspool it. Modern tech CAN
sometimes be applied ... read the tape from the non-mag side
just before it's exposed by spooling. The original will be destroyed
but a good copy can still be obtained.

Ancient hard disk assemblies - the kind where you had a dozen or
so big platters in a removable plastic case (common in biz/govt
apps in the 70's and earlier 80's) might be in such bad shape
that you can never spin them up again. Static scans with high
sensitivity probes would be necessary. VERY time/$$$
consuming. And then HOW did company-x format data for
THAT particular model of box ? How did department-X
have their comp guys sub-format their info ? "It Works"
was the order of the day from the 50s through Y2k.

If you want permanence, parchment and papyrus and clay
tablets are the way to go. We find more good records from
Babylon than from the Gemini program (which, IMHO, was
more important than the Apollo program - more things of
relevance explored).

Alas the sci/tech/culture/idea curves that so radically
differentiated the latter 20th century from the previous
50 is exactly where the problems appear. This is the
era most valuable for historical research, and yet the
most ephemeral info-wise.