Memristor cross bar arrays for faster AI neural nets and math?

[Jan Panteltje]

Source:
University of Massachusetts Amherst
Summary:
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm

bit like our neural nets...

[Jeroen Belleman]

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If
you accept that, then so are capacitors and inductors!

Jeroen Belleman

[Jan Panteltje]

On a sunny day (Mon, 18 Mar 2024 11:02:33 +0100) it happened Jeroen Belleman

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
quote from that llnk:
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation,
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

[Don]

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation

,

Post by Jan Panteltje
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?

Danke,

[John Larkin]

Post by Don

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation

,

Post by Jan Panteltje
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?
Danke,

Sure.

[john larkin]

Post by Don

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation

,

Post by Jan Panteltje
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?
Danke,

Sure.

Here are some:

https://www.dropbox.com/sh/zwv8c15rh5t791f/AADT2pjOBm1cONXXPe3I_FI8a?dl=0

When I was a kid, IBM made news by announcing that they had reduced
the price of CPU core memory to a million dollars per megabyte.

A million dollars was a lot of money back then.

[Jeroen Belleman]

Post by Don

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation

,

Post by Jan Panteltje
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?
Danke,

Memory cores have a very wide hysteresis. You need a fair bit of
current to magnetize them, and when they do, they go straight into
saturation and stay there when the current is removed. You have to
reverse the current to magnetize them the other way and again will
flip the whole way.

That's why they were useful as memory. As inductors, not so much.

Jeroen Belleman

[Don]

Post by Jeroen Belleman

Post by Don

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation,
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?

Memory cores have a very wide hysteresis. You need a fair bit of
current to magnetize them, and when they do, they go straight into
saturation and stay there when the current is removed. You have to
reverse the current to magnetize them the other way and again will
flip the whole way.
That's why they were useful as memory. As inductors, not so much.

Well said! This take on the Two-Capacitor Paradox ironically illustrates
intrisic inductance:

Learning from the Two-Capacitor Paradox: Do Capacitance and Inductance Exist?
<https://www.comsol.com/blogs/learning-from-the-two-capacitor-paradox-do-capacitance-and-inductance-exist>

Danke,

[Jan Panteltje]

On a sunny day (Mon, 18 Mar 2024 12:37:02 -0000 (UTC)) it happened "Don"

Post by Don

Post by Jan Panteltje

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a
memristor, can complete complex, scientific
computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If

If you can do matrix computations with it why not?
"When organized into a crossbar array,
such a memristive circuit does analog computing by using physical laws
in a massively parallel fashion, substantially accelerating matrix operation

,

Post by Jan Panteltje
the most frequently used but very power-hungry computation in neural networks
"

Post by Jeroen Belleman
If you accept that, then so are capacitors and inductors!

Well you could store analog info in CMOS too, even in capacitors.
Inductors? not so sure, not so easy for a long time?

Does core memory qualify as inductors?

No, that is magnetic storage.
You could use a stepper motor (also uses an inductor) to set a value to to some thing too.
Memory is not that hard.
Mechanical, old 78 rpm records store even music.
Books, printing.
Processing it more difficult, but we can read the books and do the math described.

[Liz Tuddenham]

Post by Jan Panteltje
Mechanical, old 78 rpm records store even music.

They were also used for data. The WWII transatlantic telephone
scrambler system used "one-time records" (like a one-time pad) to
synchronise the encoding and decoding of the audio bands.

(I'm not sure where I came across that, but I think it could be hidden
away in either the 'cdvandt' or the Crypto museum websites.)

[Jan Panteltje]

On a sunny day (Tue, 19 Mar 2024 08:55:59 +0000) it happened

Post by Liz Tuddenham

Post by Jan Panteltje
Mechanical, old 78 rpm records store even music.

They were also used for data. The WWII transatlantic telephone
scrambler system used "one-time records" (like a one-time pad) to
synchronise the encoding and decoding of the audio bands.
(I'm not sure where I came across that, but I think it could be hidden
away in either the 'cdvandt' or the Crypto museum websites.)

Some googling found it:
https://www.cryptomuseum.com/crypto/usa/sigsaly/index.htm

Some setup!

[Liz Tuddenham]

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device,
called a memristor, can complete complex, scientific computing tasks
while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If
you accept that, then so are capacitors and inductors!

I don't see the problem. Switches, relays, beads on a string can all be
computing devices - and not all of them have 'memories'. Pots have been
used as computing elements on the front panel of analogue computers for
years, so what is the difference between them (set by hand) and a
memristor (set electronically)?

...and yes, capacitors are definitely computing devices: Blumlein/Miller
integrator.

[Jeroen Belleman]

Post by Liz Tuddenham

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device,
called a memristor, can complete complex, scientific computing tasks
while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If
you accept that, then so are capacitors and inductors!

I don't see the problem. Switches, relays, beads on a string can all be
computing devices - and not all of them have 'memories'. Pots have been
used as computing elements on the front panel of analogue computers for
years, so what is the difference between them (set by hand) and a
memristor (set electronically)?
...and yes, capacitors are definitely computing devices: Blumlein/Miller
integrator.

Well, then so is any electronic component.

I'm old enough to have played with analog computers. They were
fun to solve differential equations with. Not necessarily
electronic, either. I've also done it with pneumatic stuff,
bellows, nozzles and flapper valves, and with water levels in
vertical pipes. Those were real fun too. You could *see* what
was going on without instrumentation.

It's been quite a while since I last did that, though. These
days, everything is electronic.

Jeroen Belleman

[Liz Tuddenham]

Post by Jeroen Belleman

Post by Liz Tuddenham

Post by Jeroen Belleman

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device,
called a memristor, can complete complex, scientific computing tasks
while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

I have an issue with calling a memristor a 'computing device'. If
you accept that, then so are capacitors and inductors!

I don't see the problem. Switches, relays, beads on a string can all be
computing devices - and not all of them have 'memories'. Pots have been
used as computing elements on the front panel of analogue computers for
years, so what is the difference between them (set by hand) and a
memristor (set electronically)?
...and yes, capacitors are definitely computing devices: Blumlein/Miller
integrator.

Well, then so is any electronic component.
I'm old enough to have played with analog computers. They were
fun to solve differential equations with. Not necessarily
electronic, either. I've also done it with pneumatic stuff,
bellows, nozzles and flapper valves, and with water levels in
vertical pipes. Those were real fun too. You could *see* what
was going on without instrumentation.

Whiffle tree?
Used in player pianos and organs for converting binary pneumatic signals
into analogue movement. (...and previous to that, for summing the
tractive effort of a team of horses.)

[John Larkin]

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

Most universities now have a team of publicists that prowl the
hallways for miracles to announce. Rags like Sciencedaily need input.

This one is even more fun:

https://www.sciencedaily.com/releases/2024/03/240314122109.htm

"The device produces energy outputs exceeding 100 volts"

Check out their source, cell.com.

[Jeroen Belleman]

Post by John Larkin

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

Most universities now have a team of publicists that prowl the
hallways for miracles to announce. Rags like Sciencedaily need input.
https://www.sciencedaily.com/releases/2024/03/240314122109.htm
"The device produces energy outputs exceeding 100 volts"
Check out their source, cell.com.

I'm surprised they don't claim you can charge your phone with it.

Jeroen Belleman

[john larkin]

On Mon, 18 Mar 2024 17:29:11 +0100, Jeroen Belleman

Post by Jeroen Belleman

Post by John Larkin

Post by Jan Panteltje
University of Massachusetts Amherst
A team of engineers has proven that their analog computing device, called a memristor, can complete complex, scientific computing tasks while bypassing the limitations of digital computing.
https://www.sciencedaily.com/releases/2024/03/240314145325.htm
bit like our neural nets...

Most universities now have a team of publicists that prowl the
hallways for miracles to announce. Rags like Sciencedaily need input.
https://www.sciencedaily.com/releases/2024/03/240314122109.htm
"The device produces energy outputs exceeding 100 volts"
Check out their source, cell.com.

I'm surprised they don't claim you can charge your phone with it.
Jeroen Belleman

The top hat on the duck turns out to be the critical element.

https://www.amazon.com/Forum-Novelties-Retro-Happy-Drinking/dp/B00GYDDQ8U