Graphene-Based Memristors Inch Towards Practical Production (phys.org) 12
Longtime Slashdot reader Baron_Yam writes: Memristors are the long-sought 4th fundamental circuit element. They promise analog computing capability in hardware, the ability to hold state without power, and to work with less power. A small cluster of them can replace a transistor using less space. Working and long term storage can blend together and neural networks can be implemented in hardware -- they are a game-changing innovation. Now, researchers are getting closer to putting these into production as they can now produce graphene-based memristors at wafer scale. "One of the key challenges in memristor development is device degradation, which graphene can help prevent," reports Phys.Org. "By blocking chemical pathways that degrade traditional electrodes, graphene could significantly extend the lifetime and reliability of these devices. Its remarkable transparency, transmitting 98% of light, also opens doors to advanced computing applications, particularly in AI and optoelectronics."
The findings have been published in the journal ACS Advanced Electronic Materials.
The findings have been published in the journal ACS Advanced Electronic Materials.
But will it be practical? (Score:2)
We already had memristors, that's what 3D XPoint was. Far from being a game-changing innovation, they flopped hard. Too expensive to replace long-term storage, too slow to replace short-term storage, nobody bought them and both Micron and Intel abandoned them.
Now, this article seems to be about a new type of memristor that's supposedly better in some way, but unless they can either match NAND cost with better performance, or match RAM performance at a better cost, nobody will buy it just like previous comme
Re:But will it be practical? (Score:5, Insightful)
>We already had memristors, that's what 3D XPoint was.
You do not seem to know anything about memristors.
3D XPoint was not made of them, and had none of their features other than persistent state without power.
the ability to hold state without power (Score:3, Funny)
Texas Republicans?
Inching? (Score:1)
Don't you mean a few NANOMETERS?
Re: (Score:2)
That's certainly more accurate. From theory in 1971 to a single physical one in the early 2000s, and here we are yet another couple of decades on and it's now "well, we can make them at some kind of scale but they're pretty fragile".
Still, it took over two thousand years from the first steam engine to one that could actually do useful work.
It's funny... (Score:3)
...my gut reaction, when reading this story about advances in computer hardware, is anxiety.
20 years ago my gut reaction would have been mild interest and "gee, that's neat".
Re: (Score:2)
Why anxiety? I don't see anything in TFS or TFA that warrants it.
There's a mention in TFA about "mimic[ing] the synaptic functions of the human brain" but it seems to me that has been done already.
Re: (Score:2)
There's a mention in TFA about "mimic[ing] the synaptic functions of the human brain" but it seems to me that has been done already.
In theory, memristors would be more energy-efficient and faster for neural nets since they are massively parallel.
Re: (Score:2)
Thanks for that. Sure, it may be a game-changer but it's not anxiety-inducing as far as I can see. Just a more efficient way to do things that are being done now.
Re: (Score:2)
it's not anxiety-inducing as far as I can see.
Indeed. I feel no anxiety about memristors.
But I'm a techno-optimist, so I rarely feel anxiety about anything.
Just a more efficient way to do things that are being done now.
With neural nets, more energy efficiency, density, and parallelism means bigger models and more training, which leads to more capability.
A description floating around the web (Score:1)
"A memristor is like a water a pipe that changes diameter with the amount and direction of water that flows through it. If water flows through this pipe in one direction, it expands (becoming less resistive). But send the water in the opposite direction and the pipe shrinks (becoming more resistive).
Further, the memristor remembers its diameter when water last went through. Turn off the flow and the diameter of the pipe "freezes" until the water is turned back on. That freezing property suits m