Researchers Claim Robust 'Universal Computer Memory' Breakthrough (lancaster.ac.uk) 123
Lancaster University has announced a "universal computer memory" breakthrough combining the fast, low-energy storage of DRAM memory with the robustness of flash memory. They're now envisioning ultra-low energy consumption computers which would never need to boot up -- and can "instantaneously and imperceptibly" slip into an energy-saving sleep mode.
Long-time Slashdot reader Hrrrg pointed us to this announcement: A U.S. patent has been awarded for the electronic memory device with another patent pending, while several companies have expressed an interest or are actively involved in the research. The inventors of the device used quantum mechanics to solve the dilemma of choosing between stable, long-term data storage and low-energy writing and erasing... [Specifically, "by exploiting the quantum-mechanical properties of an asymmetric triple resonant-tunnelling barrier."]
Physics Professor Manus Hayne of Lancaster University said, "Our device has an intrinsic data storage time that is predicted to exceed the age of the Universe, yet it can record or delete data using 100 times less energy than DRAM."
The announcement predicts the technology could reduce peak power consumption in data centers by 20%.
Long-time Slashdot reader Hrrrg pointed us to this announcement: A U.S. patent has been awarded for the electronic memory device with another patent pending, while several companies have expressed an interest or are actively involved in the research. The inventors of the device used quantum mechanics to solve the dilemma of choosing between stable, long-term data storage and low-energy writing and erasing... [Specifically, "by exploiting the quantum-mechanical properties of an asymmetric triple resonant-tunnelling barrier."]
Physics Professor Manus Hayne of Lancaster University said, "Our device has an intrinsic data storage time that is predicted to exceed the age of the Universe, yet it can record or delete data using 100 times less energy than DRAM."
The announcement predicts the technology could reduce peak power consumption in data centers by 20%.
Re: (Score:2)
Me too. I can only remember one case where the computer slowed down considerably due to swapping, and it apparently was either a bug in Firefox, or some script that started allocating huge amounts of memory.
The only cases where I need more memory are 1) virtualization and 2) compiling a big chunk of code with lots of parallel processes - and I don't tend to do that on my personal computer.
Re: (Score:2)
The situation is much better than it used to be because we all have lots of cheap ram now, and because swapping is therefore usually a sign of a runaway process which is on its way to death by OOM killer. But if you've literally never experienced a swap-related slowdown then you must have just fallen off the turnip truck yesterday. These days I just don't enable any because once you start swapping, performance goes tits up. I'd rather just get to OOM and have a process die sooner than later. Also, I don't w
Ah, progress (Score:5, Funny)
All well and good until, like with razor blades, someone starts using a quintuple resonant-tunnelling barrier.
Re: (Score:1)
Re: (Score:2)
All well and good until, like with razor blades, someone starts using a quintuple resonant-tunnelling barrier.
As predicted [theonion.com]
And, for those too young to remember, this joke started with the 1975 SNL fake ad for the Triple-Track Razor [jt.org] that ends:
Announcer: The Triple-Trac. Because you’ll believe anything.
Not WonderMemory again (Score:3)
And "data storage time that is predicted to exceed the age of the Universe" is just exceptionally stupid. Sure, maybe the bit has that property if it is made from Iron or another really long-lived material, but "data storage" does necessarily include the ability to access it and then you are at the typical semiconductor lifetimes for 20-50 years. I would expect a Physics professor to at least have a basic understanding of that. Also, a Physics professor should definitely know that we have no firm data on how long the universe is going to exist, just several competing possibilities.
The while thing is very likely to land on the very large heap of similar "revolutionary", but ultimately unworkable technologies anyways. The grander the announcement, the smaller the actual substance.
Re: (Score:2)
Re:Not WonderMemory again (Score:4, Funny)
I just throw all my data into a black hole and hope that someone develops a way to read it back out using hawking radiation when it winks out in 20bil years.
Re: (Score:3)
Ahh, write-only memory. I have infinite amounts of that already, and you wouldn't believe the speeds I can achieve writing to it.
Re:Not WonderMemory again (Score:4, Interesting)
This depends on whether you accept "eternal inflation" or which other theory. The "heat death" that thermodynamics made popular isn't really valid...we just don't know which alternative *is* valid. (And with *some* of them it could be far enough away that the heat death of everything within our light-cone could happen first.)
There's some evidence, however, that the "big rip" is a probable end...and in that case while everything comes to an end, it doesn't happen at the same time...though you won't be able to tell because space will have stretched so far that you can't talk to anyone/thing else. But the smaller something is, the longer it will be able to hold together. (I don't really think that's what's going to happen, but that's just my opinion. And there is *some* evidence that it's the expected end.)
Re: (Score:2)
There's some evidence, however, that the "big rip" is a probable end...
For your entertainment, some pipe smoking ideas:
Spacetime and Electromagnetism are what is left of energy as a result of "condensing" energy into matter.
Without matter, there is no space and no time, which means there can not be a "big rip".
Ultimately, the Universe will "dry up" into pure energy (so no spacetime nor electromagnetism) and the Big Bang will occur again. Total erasure of all previous information and a fresh new start.
That is assuming that protons and neutrons "decay".
Re: (Score:2)
Non-volatile data retention of at least 10^4s
So about 3 hours. Not quite as good as current Flash memory. I have no idea where that other figure comes from.
Re: (Score:2)
I think the "at least 10^4s" may be what they actually are confident they have amply tested on the fabricated cell, versus the theoretically modeled retention. They had some unexplained drift in the read currents they have to look into. I gave up VLSI after college, so I don't know if the number of layers they are using is excessive these days, but it looks promising if it is verified and a path to industry scale fabrication is found.
Re: (Score:3)
For the past 20 years I've been reading on Slashdot about "breakthroughs" in MRAM and how it'll lead to bootless computers within the next 5 years. I predict when such computers do arrive they will be powered by Mr Fusions.
Re: (Score:2)
We could have bootless computers now if we really wanted to. Tech like Intel Optane is as fast as RAM used to be, so for a lot of software there isn't all that much point even loading it into RAM any more.
The problem is that it would be a huge paradigm shift and no-one is willing to make it yet.
Re: (Score:2)
And "data storage time that is predicted to exceed the age of the Universe" is just exceptionally stupid. Sure, maybe the bit has that property if it is made from Iron or another really long-lived material, but "data storage" does necessarily include the ability to access it and then you are at the typical semiconductor lifetimes for 20-50 years. I would expect a Physics professor to at least have a basic understanding of that. Also, a Physics professor should definitely know that we have no firm data on how long the universe is going to exist, just several competing possibilities.
10^14 year commentary in paper is explicitly about stability of trapped charge not physical real world implementations.
"The intrinsic InAs FG is isolated from the InAs channel by a 15-nm AlSb barrier, while double InAs quantum wells (QWs) with triple AlSb barriers serve as a resonant-tunnelling barrier between the FG and the n-doped InAs CG. Hence, in our devices the electrons stored in the InAs FG are isolated by the anomalously-large conduction-band discontinuity with AlSb, a charge-confinement system tha
Re: (Score:2)
10^14 year commentary in paper is explicitly about stability of trapped charge not physical real world implementations.
I realize that. I am not criticizing the paper. I am criticizing the statement made by that professor as lie by misdirection. And it is.
Like with many other amazing announcements (Score:1)
let us know when we can buy it in the store. If ever.
Re: (Score:3)
But when they do that, people would claim it's advertising, and not news.
You'll Never Know Whether Your Data Is "Deleted" (Score:3)
Re: You'll Never Know Whether Your Data Is "Delete (Score:4, Interesting)
But encryption keys stored in ram mostly go poof when you pull the plug... Not so when you introduce nonvolatile ram...
Re: (Score:2)
Data does survive a warm reboot, and the amiga even used this feature to have a reset-survivable ramdisk that you could boot from.
Modern systems tend to be so bloated that all the memory has been overwritten at least once by the time the system boot, but if you were to load up a very light weight system and scan the remaining memory you would find remnants of the previous boot.
Re: (Score:3)
Not really, and in fact that is the basis of the cold boot attack. Hit the reset button and boot a minimal Linux distro that looks for encryption keys in RAM. Or even just pull the RAM sticks and put them in another machine. A bit of freezer spray helps them retain the data while being transplanted.
That's why encrypted RAM is such a great thing. Really glad AMD started shipping that. The crypto key is stored in a physically protected part of the CPU and forcibly wiped at reset.
Re: (Score:3)
Not just SSDs. Capacitance power has had that same effect even without intention. And some modules have had it built in on purpose. It doesn't last forever, but an hour is often enough for the intended use. Better capacitors and less leaking in the circuit should let RAM hold it's charge for days. I haven't heard of anyone designing this to handle anything more than power fluctuations, but they certainly could have even in the 1970's.
Re: (Score:1)
WTF are you talking about? That makes zero sense. Are you one of those tin foil hat types? Oh, I just read your sig. I guess you are. Carry on.
Re: (Score:2)
Re: (Score:2)
@dryriver Don't get riled up! 110010001000 comes to Slashdot to write troll posts --- not to contribute in any meaningful way.
That would become apparent to you if you check what else he has posted...
Re: (Score:2)
Lovely innovation, except that when you pull the plug on current computers the RAM gets wiped clean.
Not immediately. Successful cryptographic keys recovery attacks have been carried out by retrieving them from RAM in computers that had been recently unplugged.
Re: (Score:3)
That doesn't necessarily work if the device does "usage leveling". I've never heard of RAM doing that, but flash memory does. And disks (used to?) keep the old file on the disk, but just drop the pointers to it as used storage. If you write the program, you can ensure that the existing file is rewritten, but most programs seem to use a working copy so that if you abort without closing at least the prior version is recoverable. Some of them even rename to old file xxxxx.back or some such.
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
Not much of a claim - flash wears out pretty quickly when written often.
There's a good Barry Allen [wikipedia.org] joke in here somewhere ...
I tried to read the article (Score:4, Interesting)
Re: (Score:2)
Any research who claims that they have a "US Patent" as an accomplishment is probably a scam. Anyone can get a US patent. You just need to pay the filing fee.
Re: (Score:3, Informative)
Re: (Score:3)
One wording error in your patent, and you slam into an existing patent, or someone can still steal your idea by changing it a little bit.
Yes, like adding "with a computer" or "on a network" and filing for a new patent... Isn't that the majority of new patents?
Re:I tried to read the article (Score:4, Insightful)
Yes, companies hire lawyers to do this. It's still stupidly easy compared to the past. You don't even have to prove that your idea works, you don't need a model, you don't even need to worry too much about conflicting patents as long as no one notices. The USPTO is not doing the necessary background work here to verify that the patent is a new and novel idea. We have a wall at work with a lot of patents on it; almost all of them are essentially bullshit, ideas that have been common place in the past only this time they're being used in a slightly new context.
Everyone basically knows patents are pointless except for one thing. You need an arsenal of B.S. patents so that you can defend yourself against the competition when they attack you with some B.S. patents. The larger companies actually band together to cross-license patents while keeping out the smaller companies by demanding high fees from them.
Re: (Score:2)
If it's that difficult then why do people keep getting "on a computer" patents or stuff like the infamous Apple "rounded corners"?
Re: (Score:2)
all types of memory in use now went through early development stages even more primitive than this, and yes it takes many years, sometimes many decades, to get to tech in actual use in devices
MINE (Score:1)
Bable (Score:1)
Had to do something to jack the price (Score:2)
Re: (Score:2)
You're seriously complaining because a new, superior technology will (probably) cost more than existing, inferior technologies?
You realize that is precisely how market-based innovation works, right? Nobody in their right mind would spend man-centuries of labor and millions of dollars to develop better technology and then pass the benefits on to you for free. Without that incentive, you'd still be feeding punch cards into your university's ENIAC.
So... Another Non-Volatile Memory Tech (Score:2)
So it's fast non-volatile memory. How does this compare to other fast non-volatile memories like phase-change RAM (3d Xpoint/Optane) or MRAM?
Re: (Score:3)
Where did you see the claim it was fast? Maybe I missed it, but all I saw was extremely stable and requiring relatively low energy. I actually was going to comment that they don’t seem to talk about the speed of access at all.
Re: (Score:2)
From the summary:
combining the fast, low-energy storage of DRAM
No idea if the article mentions speed.
Re: (Score:2)
Scanning the actual paper (https://www.nature.com/articles/s41598-019-45370-1) there are definitely some questions. For example, they operate at 2.6V, which is rather high for high speed memory. The higher the voltage the longer transitions take, because we can only make the slew rate so high. Most DRAM operates at 1.8 or 1.6V.
They also don't do any comparison to SRAM, which is a shame because it could be useful for ultra low power applications that currently make use of it. SRAM is more expensive and slowe
Re: (Score:2)
My guess is that if it ever escapes the lab, it will be priced into the stratosphere as they figure out how to recoup the $10 billion in investment it takes to mass produce it and then how hardware makers can extract maximum profit from it.
Optane has been commercially available for a couple of years now, and yet its barely available as a product. There were at least initial production yield issues, but there were probably also MBA spreadsheets that said its speed/durability properties weren't enough to all
No boot == No brains (Score:2)
Until the operating systems and other software are perfect, there's going to be a need to boot and reboot the machines.
Re: (Score:1)
I'll believe it when I can buy it (Score:4, Insightful)
Until then, it's probably just noise.
What happened to holographic storage? (Score:1)
Only Unix can do that (Score:2)
They're now envisioning ultra-low energy consumption computers which would never need to boot up
Using what operating system? Nothing by Microsoft is reliable enough to never require a reboot.
Re: (Score:1)