Slashdot Mirror
Buckminsterfullerene Strikes Again - Nanotube RAM
putaro writes "Nanotube based RAM, under development by Nantero, promises to deliver densities of over 1 terabit per cm^2, is non-volatile and faster than current DRAM. The Economist has a nice story. Forget about just kicking DRAM's and FLASH's butt, is this finally the end of magnetic storage as well?"
A new type of computer memory uses carbon, rather than silicon
WAITING for a computer to turn on is a nuisance. That is why manufacturers have been trying to create "non-volatile" memories. These would be fast, like the random-access memory (RAM) chips that are currently used for often-accessed memory, but they would also continue to store information even without power, like hard drives, which are too slow to use except for long-term storage.
Several technologies have been competing to become the standard for fast, non-volatile memory. The best known is magnetic RAM, which IBM and Motorola are touting. Others are based on polymers or on strange-sounding metal alloys called chalcogenides that change shape when an electric charge is applied to them. But there is now a new entrant to the field: carbon.
Carbon comes in many forms. Diamonds and graphite are two of the most familiar ones. A less familiar variety is the nanotube, also known as a "buckytube" after Richard Buckminster Fuller, whose geodesic domes have a framework similar to the arrangement of the atoms in a nanotube. Nanotubes consist of a cylindrical array of carbon atoms whose diameter is only about 1 nanometre (a billionth of a metre). If Nantero, a firm based in Woburn, Massachusetts, proves correct, such tubes will soon be an integral part of computer memories.
Nantero's memory chips consist of billions of nanotubes, each a few hundred nanometres long, suspended from a silicon wafer. Another wafer sits about 100 nanometres below the first. Because the nanotubes that Nantero uses conduct electricity, a small electric charge at one point on the second wafer will draw several dozen nanotubes towards it. Once they are there, they stay there. That is because they are bound by Van der Waals forces--intermolecular bonds that do not depend on external power for their maintenance. An additional application of current, however, will release the nanotubes. This means that a group of a few dozen nanotubes can act as a memory element, storing a single bit (either a one or a zero) of the binary code that computers use to operate. If the connection between the wafers is live at a particular point, the bit represented is a one. If not, it is a zero.
If nanotubes were not so small, this would not be a big deal. Because they are, though, Nantero's technology can already achieve a data density considerably higher than existing RAMs. And because the wafers are so close together, those data can move rapidly from place to place. Nantero's new memory can read or write a bit in as little as half a nanosecond (billionth of a second). The best RAM chips, by contrast, need ten nanoseconds to perform a similar operation.
At the moment, Nantero has only a working prototype. But the firm aims to have memories on the market within a year. It thinks it will be able to tool up for commercial production quickly, because the fabrication technique it uses, though novel, relies on standard semiconductor-making technology.
The main difficulty faced by others who have tried to go down the buckytube route is getting the tubes to align with each other when they are hung from the first wafer. Until now, the approach has been to try to grow all of the tubes in the correct orientation to start with. But Nantero's founders came up with a simpler, if less elegant, solution. They use established lithographic techniques to get rid of tubes that are pointing in the wrong direction by zapping them with an electron beam. That leaves only those that are hanging down towards the opposite wafer.
Though the recent chip is certainly impressive, the reason for getting excited about Nantero is not so much the present as the future. Unlike silicon, which is pushing against its physical limitations, carbon-nanotube technology is in its infancy. Greg Schmergel, Nantero's boss, says that within the next few years the firm's engineers may be able to achieve data densities of a trillion bits per square centimetre (more than 1,000 times that available on existing RAM) and it will be possible to read those memories 100 times faster than can be done at the moment. The days of silicon-based memory may be numbered
Something else I can stare longingly at on newegg while knowing full well i'll have to sell my wife and 2 pints of plasma to actually buy it...
well, it's nothing one behind the ear wouldn't cure
--
Need a calculator?
Here's a little details that pretty much summarize the docs:
How it works. Nantero's memory chips consist of billions of nanotubes, each a few hundred nanometres long, suspended from a silicon wafer. ... This means that a group of a few dozen nanotubes can act as a memory element, storing a single bit (either a one or a zero) of the binary code that computers use to operate. If the connection between the wafers is live at a particular point, the bit represented is a one. If not, it is a zero.
Speed. Nantero's new memory can read or write a bit in as little as half a nanosecond.
Availability. At the moment, Nantero has only a working prototype. But the firm aims to have memories on the market within a year.
Hurdles. The main difficulty faced by others who have tried to go down the buckytube route is getting the tubes to align with each other when they are hung from the first wafer. Until now, the approach has been to try to grow all of the tubes in the correct orientation to start with. But Nantero's founders came up with a simpler, if less elegant, solution. They use established lithographic techniques to get rid of tubes that are pointing in the wrong direction by zapping them with an electron beam. That leaves only those that are hanging down towards the opposite wafer.
--
Error 500: Internal sig error
This guy is right...
"Ford," he said, "you're turning into a penguin. Stop it."
Get some people working on power supplys and rod logic, and dimond age here we come.
Someone should be able to make a decent mp3 player with this stuff...
Is there anything nanotubes CAN'T do? It seems like new uses are being discovered for them every day.
Though the idea of using a material that burns when exposed to a camera flash, for storage, is a little unnerving... Anyone know how they plan to address that and other problems/inherent properties of nanotubes?
using namespace slashdot;
troll::post();
"At the moment, Nantero has only a working prototype. But the firm aims to have memories on the market within a year."
A lot shorter than ten years, hopefully. Though I'm skeptical we'll see them commercially available within a year...
I hope it would be compatable with existing memory systems, though. It would be nice to just swap out existing RAM for a NanoRAM module and get an instant performance and capacity boost (Providing the controllers don't become an issue, but that's where AMD's 64-bit chip and it's built in memory conntroller come in!).
=Smidge=
1 terabit per cm^2...
I can write a byte's worth on a cm^2 piece of paper; Just repeat many times and stack; when measuring measure from above.
I guess it means lots of space (for data) in a small amount of (spatial) space. 120something GB in the space of your thumbnail.
Bitter... No, not me.
"Learning is not compulsory... neither is survival."
--Dr.W.Edwards Deming
How is that informative? "I hope it would be compatable with existing memory systems, though. It would be nice to just swap out existing RAM for a NanoRAM module and get an instant performance and capacity boost"???
Maybe you didn't notice - it said TERABIT/cm^2. Your current system probably can't handle more than 2Gig of RAM - let alone hundreds and hundreds of gig. Hell, your BIOS may not even be able to handle a HD that large.
I mean, really. 100x faster, and >1000x the storage. Think about that for a minute. Who gives a fuck if you have to toss your HUGE SLOW FUCKING SYSTEM and buy a new one?
If they do pull this off (and I think they're blowing smoke), it will make today's computers look like the vacuum tube machines of yore.
Hope to drop in upgrade... Come on.
Given the description of how it works, I wonder if it will be inherently less durable against electric shock than current hardware. We've heard the advantages, it'll be interesting to hear what the disadvantages might be. Things like failure rate and recovery methods come to mind. Definetly worth watching though!
Ryan Fenton
For those who are interested, the Nantero's technology is based on earlier work in the lab of Charles M. Lieber. The original paper was published in the journal Science. Rueckes et al, Science, Vol 289, P. 94. Rueckes went on to found Nantero.
The original experiment worked as follows:
The original experiment was done with bundles of carbon nanotubes. In principle, the concept should work at much higher densities for single nanotubes, but the technology still has hurdles to cross. Currently, the tubes conduct because ropes of tubes are likely to contain both semiconductor type and metal type tubes. Since metal type tubes are fantastic conductors, having even a few of them in a rope will allow a device to work. However, when one crosses the threshold to single nanotubes, the device will only work if the tubes are metal type. Hence, an important problem will be finding a way to produce only metal type single walled nanotubes. Currently, carbon nanotubes are produced in a mixture of semiconductor type and metal type nanotubes. It's difficult to control that property because it depends sensitively on the way the sp2 bonds on the nanotube sidewall line up, something that no one yet knows how to control.
Anything that dense and that much faster than existing memory technologies will need a different memory controller. No getting around it.
The fact that AMD built a memory controller into the Opteron is not necessarily something to be happy about. On the one hand, it greatly reduces latency of memory reads/writes, on the other hand you can't upgrade the memory speed beyond what your entire CPU supports -- you have to upgrade your entire CPU. Which means AMD has to redesign the CPU to take advantage of faster (or different types of) memory. And Opterons aren't that cheap yet...
you can't upgrade the memory speed beyond what your entire CPU supports -- you have to upgrade your entire CPU. Which means AMD has to redesign the CPU to take advantage of faster (or different types of) memory. And Opterons aren't that cheap yet...
Not so. The opteron has 3 hypertransport busses which can be connected to alternate memory controllers - the onboard one is then disabled. What I want to know is whether AMD plans to maintain separate part numbers for each speed/controller combo, or if they're just going to band them, with higher clocked Opterons getting faster memory.
"We returned the General to El Salvador, or maybe Guatemala, it's difficult to tell from 10,000 feet"
I'll be able to start a Java applet in Mozilla running on top of KDE.
Just kidding, in fact I just want to run Nautilus.
sgis ddo ekil t'nod i
You couldn't even track it by user at the OS level (user a has memory x and y allocated, so user b can't use that.) because I could still boot it into a different OS through a removable drive...
Of course, you could just eliminate all caches of keys or passwords... But do you really want to have to re-enter your slashdot password everytime you hit refresh, or click on a link to the comments page, or click to read a reply?
Maybe the solution would be to specify a certain area of RAM that would get initialized on power-up (be it a reboot or just waking up from an NVRAM suspend), and get apps to put any sensitive information in that area... Which would probably require additions to your favorite OS's API, in addition to new versions of a lot of apps...
Just thinking 'out loud' here... Anybody else thought about this?
The subject says it all....
What's under yellowstone?
This is how big their protoptype is from their website:
"Dr. Thomas Rueckes, Chief Scientific Officer
and Co-Founder said, "This gets around
the problem that nanotubes cannot reliably be
grown in large arrays. At the end of our
process only the nanotubes in the correct
positions are remaining. This process was
used to make a 10Gb array now, but could easily
be used to make even larger arrays--
the main variable now controlling the size is the
resolution of the lithography equipment."
well... if your name is minime disregard this ;-)
In Canada, we don't fancy things like socks
Oops, sorry. I forgot cases, where usually there is not enough of power sockets and spaces for additional hard-drives. And don't forget floppy drives - they are still here, in most PCs I see in the store.
I can easyly imagine to see, in a year or two, a PC with several TB of nanotube-based RAM and 1.44MB floppy drive, all connected to AOL with 56K modem.
Less is more !
It's a very cool idea, but I'm wondering why they didn't mention these issues. Is it an unmentioned limitation of the technology, or a limitation of the Economist's journalistic scope?
Okay, correct me if I'm wrong, but even though they say it has potential for hold a terabit/sq.cm., I fail to see how it mandates that RAM modules made with this tech hold more than 128/256/512MB like current modules do. The point is that you would have faster, non-volitile RAM that would fit into existing hardware.
:)]
Or, if you prefer, new computers that can still use the older (and likely cheaper, at least initially) silicon memory.
I don't know about you, but I certaintly don't throw out my system every time they make something better. And let's not forget the people who are so anal/paranoid about performance would pay through the nose for a 10% increase in memory speed. I'd also imagine that this non-volitile RAM uses next to no power, making it a great potential drop-in replacement for laptops and other portable devices. A little bit of software tweaking and you could basically have an instant hibernating mode.
Useful for desktops, too... not everyone can afford/think they need a UPS system, but if the power goes out, smarter software would just bring the machine back to where it was with no data loss! Sounds like a perfect nitch for a compatable drop-in replacement to me. What's the total cost to install new RAM and flash the BIOS and/or apply a patch compared to buying all new systems?
<joke>
Of course, Silicon RAM is to NanoRAM as a kumquat is to a watermelon, but that doesn't mean people are suddenly going to stop buying kumquats and switch to watermelons. I'm using this simple analogy an an insulting way of making a point because, from the way you phrased your comment, I can see your sense of humor and practicality is limited. Unfortunately, I've lost track of the point I was trying to make, but I'm sure you'll agree that both watermelons and kumquats have their season, and either or both make a fine addition to any table.
</joke>
[credits to Dr. Science for that last bit
=Smidge=
Well, maybe not a drop in ram replacement, but how about a solid-state ide/scsi drive? Memory blocks, so to speak...
Look, you're missing the point. This would represent a fundamental change in computing. 1 terabit/cm^2. Imagine having 50 GBytes on your wristwatch.
Half a terabyte on your cell phone.
As many terabytes as you can imagine on a laptop that runs for a day because it doesn't have a HD and all the RAM is NVRAM, and it's 100 times faster than your current system.
Really. Think about it. Who gives a shit if you can upgrade your current machine. Did you see the article recently about AppleII users getting together? You'd look as silly as any of them... (no offense - I've run a IIe emulator within the past 3 months, and it was fun; a lot like it will be amusing to be able to store all of silentbozo's files on my cell phone many times over)
Although your right the 1000x thing and 100x thing were potentials they expect within the next few years, right now they said it was 20x faster then the fastest memory on the market today, so it will be bigger capacity although they didn't say how much bigger, and it will be quite a bit faster.
The hardware aside, in software this memory could easily replace the hd and ram, at least on a linux system. The problems I see are that all software is currently designed to use both memory and disk as if they are two separate things... and to use one device in place of both would require a rewrite. In linux talking to hardware is done through the kernel, so changes to the kernel to "emulate" ram could get this hacked into usability fairly quickly. It could allow the amount to use as ram to be passed as a kernel parameter.
as I said buried in another thread, in the overall scheme of things this should require relatively small changes, remember we are talking about replacing disk and ram. The kernel handles access to these and can already address very large amounts of memory. The hardware would actually work as a harddrive and would have a swap partition that the kernel uses as "simulated" memory so it doesn't break existing applications that have things like say... variables. Or it could be a file on the disk in which this virtual memory structure exists and you pass the amount of "memory" you want to the kernel on boot... or course MS would fail to address this issue fast enough and finally will be left in the dust.
Of course, Silicon RAM is to NanoRAM as a kumquat is to a watermelon
No. Silicon RAM is to NanoRAM as Vacuum tubes are to Silicon. They're both there to do the same thing, it's just that nobody bothers with vacuums (except for a very few special purpose - like audiophiles) because they're old and clunky.
The point is that you would have faster, non-volitile RAM that would fit into existing hardware.
No. The point is that you could easily have so much RAM that it would make retrofitting it into a current system look like putting an spoiler on a model-T.
I'd also imagine that this non-volitile RAM uses next to no power, making it a great potential drop-in replacement for laptops and other portable devices.
Good thinking. Oh, and let's not forget you wouldn't need to spin a disk at 1000s of RPM, which uses some energy as well.
This is not an upgrade. It is a change.
... for 64-bit addressing.
If you have this wonderfully fast and compact memory, the simplest way to exploit it is to access it in a linear manner with a whompin' huge address space.
Who needs VM? -- Actually, we'll still need mechanisms to isolate processes from each other, so virtual addressing will still have a place. But not as a means to accomodate logical address spaces larger than physical address spaces.
I want a fuel-cell powered, IBM 970 Powerbook with buckytube memory and an OLED display. Never mind the power switch, I'll just refuel it every other month or so.
Where do I invest in this nanotube technology. Every 10 seconds these thigns have another use that is lighyears ahead of anything we have now. One day I'll wake up and pull off my Nanotube based dirt proof - tempeture regulated blankets, step out of bed and go to the bathroom where I turn on my nanotube fillament based lights which last 10,000 years. I'll use the nanotube based super computer inside my razor to give me the perfect shave with no razor burn. Then Ill head into the kitchen and pour myself a big bowl of nanotube-crunch...
But seriously, is there anything these thigns can't do? And where can I get a peice of the action?
It doesn't change anything. Like you said you can get this now. I work in security and ANY system that a knowledgeable person can lay hands on they OWN it. As part of my job I do security audits and I have this little stickers that say "I OWN THIS" if I can get close enough to put a sticker on it then the statement is true. I doubt that will every change, but who knows what the future holds.
Encryption: I may not agree with what you say, but I will defend your right to encrypt it...
how are they going to wire the thing? Suppose the nanotubes are grown, properly aligned and so on. How are they going to place the wires between them? AFAIK, the current technology for wiring the chips is exactly the same that puts the transistors, namely the photo-process. Obviously, this si not going to work on the scale of nanotubes.
I already leave all progs running all the time...
Luke-Jr
Well, if you use an 8 terabyte equivalence for 1 Library of Congress (the actual definition seems to be a bit slippery, and that's the first one I found on a Googling)...
.62 to .73 inches thick, but since we don't know how thick nano-ram is, let's just assume a wafer the size of an iPod).
Theoretical Nano-Ram storage capacity == 1x10^12 bits / cm^2
1x10^12 bits = 1.25x10^11 bytes = 116.4 Gigabytes
That's 0.114 Libraries of Congress per cm^2.
An iPod, according to Apple's website, is 4.1 by 2.4 inches (it's also
1in = 2.540cm
4.1in = 9.840 cm
2.4in = 6.096 cm
Let's chop a cm off each of those to account for the casing, structural bits, and soldering points that aren't actually storage space. That gives us a size of 8.840cmx5.096cm for our hypothetical nanoPod (so on a tangent, how long before some company introduces the new 'e' and starts dubbing products 'nRAM', the 'nPod', 'nTel nSide', etc?). That's a surface area of 45.049cm^2.
Given our previous determination that we can store 0.114 LoC on 1cm^2, we arrive at a figure of 5.136 LoC/i(or LoC/n for nPod, as the case may be).
"If a man hasn't discovered something he will die for, he isn't fit to live" -- MLK, Jr.
First, I have to confess that I am a Materials Engineer and not some ubergeek with a CSE degree.
But it's a definite fact that technological advances are only made possible with the precedence of metallurgical advances.
Silicon wafers today wouldn't exist without the metallurgical backing to create high purity Silicon, Aluminum, and so on.
The point being that with the discovery of the buckey ball, we are entering a new age of history. We're not there, but we're working on it really hard.
Before you toss me out as flamebait consider that each primary age of human civilization is named as a metallurgical Age: Bronze, Iron, Steel. Some might argue that we are in the Silicon Age right now. However, the impact of Silicon is not as ubiquitious as the impact of the discovery of Bronze, Iron, or Steel.
But the Buckey Ball is going to be similar in the scope of impact as Steel or Iron. Why?
- Structural Materials
- Electronics
- Optics
- Aerospace
It's a FUNDAMENTALLY new material product available for the engineers to play with.So far everybody is missing the point including the Economist article. This stuff would replace SRAM. High performance FPGAs from Xilinx and Altera are made of SRAM with refresh in the neighborhood of ten nanoseconds. This would make vast and fast FPGAs possible.
So, instead of merely replacing system RAM or storage this would replace the CPU, the memory controllers, the video card, the sound card --it would be the ultimate SoC platform.
the memory manager and the part of the device manager than manages the hard drive will be one unit!!!
so to get full potential from this we will have to redesign our OSs otherwise if we stick it in the current system it will be a fraction of its potential.
I am the Alpha and the Omega-3
Mark Brehob
As Nietsche famously said, "If you stare too long into the Abyss, 1d4 Tanar'ri of random type will attack you."
How do you keep particles of RAM out of your lungs now? I'd go with the same method.
Wow, I have to upgrade the entire CPU. Like, I get a new motherboard with 128Gbytes of nanotube memory, but I can't put bits and pieces of my old Opteron on it. Perfectly shocking. What was AMD thinking.
--When components start getting this small, the chances of having an potential error occur go WAY up. What I'd be interested to see, is what they're doing to protect against stuff like cosmic-ray bit pollution and such.
--After all, if the scale is NANO, one cosmic ray or stray electro/magnetic field can potentially screw up a lot more percentage of memory... Massive redundancy, high speed and constant bit cross-checking would seem to be a reasonable requirement for these chips.
--For just one example, look what a few scratches can do to a CDR - or worse, a DVD. If you can't read it (use it reliably XMillion times) it's basically not very useful...
.
== WolfriderV6 == I'm willing to admit that *I just might* be wrong... Are you??
Actually for most peoples desktop machines S-ATA and Ultra 320 SCSI are faster than their PCI busses (33Mhz-32bit PCI is only 133MB/s). Besides, who cares if the bottleneck moves to the physical connection, at least it is away from the slow arse hdd's of today (8+ms access time and slow max transfer vs .5ns access and fast as your bus can handle transfer). I would upgrade in a heartbeat. Plus removing on of the last important physical device would do wonders for improving the reliability of computers (the cpu and case fans are still there). To get the most out of this you might need to redesign things, but I would be willing to bet that the first uses are for retrofits, market inertia tends to work that way.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
A fullerene is a certain class of carbon molecules that have a tubular or spherical form. The Buckminsterfullerene, or "buckyball" refers to the specific spherical C60 fullerine (shaped like a football). All other non spherical molecules (eg tubular) are just referred to as fullerenes.
Well, I don't think so, actually.
/forever/. This is just one example, there are *many* more, independent of operating system (I'm unfortunately forced to use Windows for work).
You still need a bootstrap that will go reset the memory and restart the OS.
If you don't, what's going to happen is that some bug is going to manifest itself, and it's never going to go away because you have no method resetting it - if you turn the machine off, then on, the messed up driver (or whatever) comes back in exactly the same bogus state.
Think about it. I've seen bad net (VPN) drivers for windows cause my normal networking drivers to be very slow once the vpn comes down, the only cure being a restart (when it happens, which is not often).
So now, with, as you say, the death of the concept of bootstrapping, when that happens, I'd have to live with slow network drivers
You will still need the 'entire concept of bootstrapping', even if only as a cure to some other guys kernel-space bugs.
Code or be coded.