If I understand this correctly, the technology replaces transistor-like technologies, where the state of one input determines if a pathway conducts or blocks, with something called memristors/spin-valves, where the input directs current rather than ever blocking it.
I recall a talk at MIT back in the 90s where the speaker emphasised that heat generated in semiconductors is generated by the frequency with which heat is generated and then sunk, and that it would be possible to design much lower energy, lower heat electronics if ideas from linear logic were applied to gate construction, minimising the need for generate-sink. The paper's technology seems to support this thesis.
For more about linear logic, see https://en.wikipedia.org/wiki/Linear_logic - I'm afraid I don't have any references to applications of linear logic to low-power device design.
> heat generated in semiconductors is generated by the frequency with which heat is generated and then sunk
For people interested on this, RWT has a very good explanation of the way we build things now and why we do it, in their series "The stuff dreams are made of".
Something like that. The 2008 symposium about memristors talked a bit about “material implication” as a basis for more efficient logic systems. http://www.youtube.com/watch?v=QFdDPzcZwbs
The HP guys published some papers about it too, but they might be all paywalled?
I don't understand why Leon Chua keeps popping up first when memristors are discussed. Williams at HP is the person that made it a reality. Chua postulated the existence of the memristor but as far as I know made no progress on a working example.
That's a bit like giving da Vinci credit for the invention of the helicopter.
The Williams talk on youtube is really good. He discusses the element itself, its use in memory (as vast amounts of on-CPU memory) as well as the possibility of new CPU architectures based on it:
This is nothing like failing to recognize that Laplace did most of the work in Bayesian Probability. Check http://en.wikipedia.org/wiki/Memristor#Historical_timeline. The subject did not suddenly explode from nascent form to real subject due to Williams' efforts.
Summary: Without the theoretical foundation of Chua people would have gone on for much longer finding the same concept in different guises without realizing that these were really just aspects of the same phenomenon. In addition, there are those who claim priority on HP's underlying tech and there is even a paper publishing similar experimental results as far back as the late 60s which the HP team did not cite. Again, if this was due to mere ignorance then the unifying power of theory to carve a template so that experiments did not continue to be repeated blindly strengthens the value of Chua's work. There is also criticism for the currently liberal labeling of what constitutes memristive behavior and HP is implicated in trying to do a patent landgrab.
If one person's work had to be picked as the inflection point then it would have to be Chua.
It has been known for decades that laws of nature imply a fourth fundamental electronic element (besides resistor, capacitor and inductor).
>>> The Memristor is believed to be the missing 4th circuit element. Memristor is basically a charge-dependent resistor. The reason that the memristor is radically different from the other fundamental circuit elements is that, unlike them, it carries a memory of its past. When you turn off the voltage to the circuit, the memristor still remembers how much was applied before and for how long. That's an effect that can't be duplicated by any circuit combination of resistors, capacitors, and inductors, which is why the memristor qualifies as a fundamental circuit element. Memristors can be combined into devices called crossbar latches, which could replace transistors in future computers, taking up a much smaller area. They can also be fashioned into non-volatile solid-state memory, which would allow data density of about 100GB/1sqcm. This can be implemented for designing circuits with lesser components, lesser expenses and lesser power consumption. Also coupling the crossbar latch with FPGA (Field Processing Gate Array) technology, we can potentially make processors so dynamic that they would never get obsolete. Giving a practical solution to reduce the millions of tonnes of computer waste generated each year...[1]
Blaise Mouttet has written numerous papers and online comments which challenge the claims put forward by memristor proponents[1][2]. Debate surrounds whether the device created by HP is a memristor or simply a varistor[3][etc]. The heated debate has been widely publicised in the scientific community[4][etc] and is still ongoing. According to [3], Blaise has an interest in preventing a patent land-grab over these technologies on the basis that the patent claims may use unjustified science.
I also note that the 2008 symposium talk by Leon Chua made mention of Memory Capacitors and Memory Inductors[5]. Stanley Williams started the next talk by thanking Leon for the challenge posted to systems implementers to find and use memory capacitors and memory inductors. If the missing circuit component is the memory resistor (confused with the varistor?), where do memory capacitors and memory inductors fit in?
Material implication is the usual propositional logic that is used in generate-sink kinds of gate design.
I think maybe you mean strict implication? I read a paper a few years back that showed how you can use strict implication to design regular propositional logics that can house resource sensitive logics like linear logic.
"One reason the brain is so power efficient is that neural spikes charge only a small fraction of a neuron as they travel. By contrast, conventional chips keep each and every transmission line at a certain voltage all the time."
On thr contrary, the brain maintains a constant voltage on neurons, and spends a lot of power to maintain that voltage in the face of leakage current. It is just a much lower voltage. (50 mV rather than 1000 mV.)
Like Singularityhub or phys.org or sciencedaily, TRN's SOP is to credulously repeat the claims of dubious papers, and give them massively misleading titles.
The most promising approaches I've seen combine analog pattern-matching-based compression of calculation with traditional digital computation. We see this right now, where human+computer regularly outperforms either alone: the goal is to move both types of reasoning into hardware.
I recall a talk at MIT back in the 90s where the speaker emphasised that heat generated in semiconductors is generated by the frequency with which heat is generated and then sunk, and that it would be possible to design much lower energy, lower heat electronics if ideas from linear logic were applied to gate construction, minimising the need for generate-sink. The paper's technology seems to support this thesis.
For more about linear logic, see https://en.wikipedia.org/wiki/Linear_logic - I'm afraid I don't have any references to applications of linear logic to low-power device design.