Core store!
On a very small scale. And without the electrical overhead.
There could be some iron in the non-volatile soul. A team of Cornell university boffins using bismuth ferrite have discovered a way to store bits on magnets without needing electric current-based switching. Their room-temperature magnetoelectric memory device has 2-step magnetic switchability "with nothing but an electric …
PAL colour TV sets used a glass delay line to store a single scan row. I think this was probably the world's major use of delay lines. It considerably postdates LEO II. The delay line used a fundamental of 4.43MHz, so the clock rate was much too high for early computers.
"2-step magnetic switchability "with nothing but an electric field". It doesn't need voltage to turn it on."
A voltage implies an electric field, since it is a potential between 2 points and so there must be a field between those points.
The article seems to suggest that a PD is needed, but no current (other than that to establish the PD obviously).
Ramble - the photograph reminds me of how much more interesting electronics looked back in the day. A board full of 2N404s and a core store look such a lot more purposeful than a bit of plastic covered in little black bugs.
Yup, there must be some energy expenditure, i.e. current. Otherwise that ferrite thing won't change its state (if it did it with no energy expenditure, it would be unstable and not a good memory-holding physical device). Also, Landauer's Principle asserts "that there is a minimum possible amount of energy required to erase one bit of information, known as the Landauer limit: kT ln 2", i.e. 0.0178 eV at room temperature.
To complete the circle, 0.0178 eV is the energy dissipated when a charge equal to 1 electron charge falls through a potential difference of 0.0178V. Current, voltage.
It is left as an exercise for the reader to show that this is the minimum noise energy that can be detected by an audiophile using those loudspeaker cables with directional arrows on them.
Audiophiles might expect it, but real junk scientists know that the hydrogen released in gold plating causes noise at the interface. True audiophile connections must be made by bolting the unidirectional cables to the terminal posts using drawn OFHC copper bolts, and the joint must be kept in an argon atmosphere to eliminate the noise caused by cable oxidation. At 0K, so that the noise of the argon atoms hitting the cable doesn't cause signal degradation.
My worry? One day I'm going to write some stuff like this and it will end up in an audiophile magazine.
Core store not only needed a fair old write current, but the readout is destructive, so every read cycle needs to be a read-modify-write even if just a read is wanted.
The old PDP-8, which was of that era, had a read-and-clear operation (DCA) which was quick simply because it was just a destructive read, but usually with core you are reading two operands, and replacing one with a new value, so a minimum of 2 reads and 2 writes is required.
However, the time for which core could retain its state was impressive. I remember removing an ancient PDP-11 from storage (we actually wanted the power supply) where it had lain unwanted for years, and out of curiosity firing up, whereupon it executed the last program it had held, leaving off from where it had shut down. No wonder they use core in the guidance computers of ICBMs.
Bubble memory. One of the coolest technologies out, a brilliant concept, but it just didn't scale. And Esaki diodes, which were dead in the water once i2l came along - and that too lost its niche once SOS arrived.
The history of computing must be littered with more brilliant concepts that turned into dead ends, in a shorter timeframe, than anything else.
Another cool memory technology that we'll hear about from time to time over the years that will never amount to anything.
Am I becoming cynical, having heard about bubble memory as a kid and about a dozen other fabulous technologies since from whatever that holographic disc technology was to IBM's racetrack to phase change memory that was supposed to be the next big thing a decade ago. The one out of all of them that sounded the least promising, flash, is the only one that has had any impact in my day to day life.
So forgive me if I fail to be as excited as some of the rest of you seem.
Got to say that I unfortunately agree.
20 years from now we'll be seeing zilch based on this tech. Maybe, some time in the next 50 years, we'll hear of some new tech "soon to be put into production" that will base itself on some variant of this discovery, but it's in the lab right now and there's no guarantee that it'll ever be getting out in any form we can easily associate with this tech in its current form.
There are literally hundreds of stumbling blocks before this gets into a finished product, the least of them being strictly technical in nature. There will be power plays, there will be infighting, influence wars and outright patent buying with the risk that the buyer has no intent of actually producing because it harms an existing production line.
So, yeah, this sounds interesting, but there's a whole forest of knives to walk barefoot through before ever seeing a price tag on something using it.
Electronic devices in the homes around me has rendered AM /SSB Radio reception at any frequency useless due to all the noise generated by those uncountable digital devices operating 24x7x365...
Joke Alert= random electromagnetic noise has a voltage component by any of Maxwell's Field eqiations...
Q= how is the voltage component of background noise different from a programming voltage that is advertised as very low ??
IMHO= we went from ferrite everything to get the ferrite out of everything as non-linear magnetic devices were inherently unstable...RS.