Magnetic Microchips 140
Mr_Ceebs writes, "Looking at the BBC today I find a new Magnetic rather than electronic chip type. The design can raise the number of chips per cm by a factor of about 1000, with the preliminary stages of the technology. For all devices this would mean the demise of the large battery pack. " H : This is a follow-on to this morning's story on moldable magnets.
Weight (Score:1)
5,500 million transistors (Score:2)
Magnetic transistors? (Score:1)
Hehehe something to hold us over (Score:1)
IS this magnetic switching of bit just as fast as traditional electronic chips, or does it impose latency. Last, and finally all I want it to run quake 3 arena on my laptop for 7 hours at 60 FPS, then Ill be happy
Speed? (Score:2)
Yeah, but... (Score:1)
kwsNI
even better! (Score:3)
Even better, when you're not using your computer, you can stick it to the 'fridge.
George
Re:Hehehe something to hold us over (Score:1)
MagRAM? (Score:2)
Refridgerators (Score:1)
Bulky Batteries? (Score:3)
This is not entirely true, for mobile phones at least. The bulk of the power consumption is from the antenna. Your standby time could be increased greatly, but talk time will likely be uneffected. Thus, in order to talk for a reasonable length of time, batteries will probably stay the same size.
Why all the stories on magnets all of a sudden? (Score:1)
Save money on doctor bills (Score:1)
Re:Yeah, but... (Score:1)
Bubble memory is back? (Score:2)
Hrmmmmmm (Score:1)
Judg3
----------------------------------
poor wording (Score:2)
"The other big advantage is energy consumption. Electronic chips use up energy during operation, whereas a magnet does not."
Processing information but not using energy. I don't think so.
Processor or Storage? (Score:1)
I don't think that the article is exactly clear on whether or not this technology is useful for building new kinds of very dense, low power memory chips (something like a very tiny bubble memory chip) or if they can use the "magnetic" transistors to build gates, etc..
This would make a lot of sense to me if these were memory elements. This would be astonishing if they had some new way of manipulating magnetic domains and turning them into basic boolean operators. (To tell the truth, I think the author of the news report isn't very clear on what the difference is.)
nuclear magnetic resonance (Score:1)
Solar? (Score:2)
Real old tech (Score:3)
Body Power? (Score:1)
Re:Yeah, but... (Score:1)
---
script-fu: hash bang slash bin bash
A few remarks (Score:1)
One thing did bug me about this article though, which is the lack of many numbers. They gave the # of transistors they could fit into a certain space, but what about the speed, heat, and actual electricity usage (even rough estimates would be nice).
Oh well.. Something else to put in the pile of knowledge of stuff that might be cool a few years from now (can anyone say new PDA's?
Magnetic Microchips... uses thereof... (Score:3)
Controllers for space/time capsules (see Doctor Who: Wargames)
Linux port for iron filings
Compass that doubles as a Quake 3 client
Levitating trains (aka: Bullet train) with SETI@Home and distributed.net clients
Finally, for the more serious-minded, I'll be interested to know if this is specific to ferro-magnetic objects. Superconductors are also magnetic, but in a completely different way.
Re:Speed? (Score:1)
And you were worried about your floppy drives... (Score:1)
moe
Re:Magnetic transistors? (Score:3)
Re:Magnetic transistors? (Score:2)
They're not like transistors, really. They're switches alright, but they operate a little like core memory in that the held charge in those 'magnetic holes' is enough to raise the energy of the potential circuit to a point where it will make it through the gate if charged, and stop dead if not. How and if they're refreshing the memory I don't know..
Re:Speed? (Score:1)
Judg3
----------------------------------
Slow... (Score:2)
That said, this would be perfect for things like space probes like Voyager. As I understand it, the space program is one of the last remaining manufacturers of core memory. In case you're too young to have seen core memory (I am--I once worked with a man that had his own computer museum), it consists of thousands of wires making a grid, with a little magnet at each intersection.
Quantum effects (Score:1)
Re:5,500 million transistors (Score:1)
With a thousand times more transistors, memory bandwidth problems might not be such a big deal. I think on the coppermines the 256k cache takes up around ten million transistors, give or take. That means we could have a coppermine or athlon core with approximately 138 megabytes of L2 cache.
Your budget integrated system could have everything integrated into that chip. Video, sound, I/O. Is it possible to make a chip that you could just plug your keyboard, mouse, monitor into (not the chip literally, but so there is no external logic)?
Spooky (Score:1)
You know when you go to University and you look around and wonder who'll end up being the captains of industry, the Nobel prize winners and the people behind the next great innovation, or who'll be practicing their lines "Would you like fries with that?"? Well it's still a major shock when I read this story and recognized the name Dr Russell Cowburn - this guy was the person who showed me the ropes when I first went to Uni. All I can say is that if he makes a mint out of this, I wish him well! And given that Cambridge Uni generally tends towards reticence rather than early disclosure on these sort of news items, you can guarantee that this technique has seem some serious peer review already. That doesn't necessarily mean that scaling this technique up to marketable levels is easy, but it sounds like the science is well understood.
Just goes to show ... I really should studied harder in those lectures on condensed matter physics. :-)
Cheers,
Toby Haynes
Re:Did we learn nothing from Y2K? (Score:1)
I think the big thing about magnetic anything is that it degrades over time as it demagnetizes, as far as I can see optical based media doesn't suffer from this. (Except for those deliberately degrading DVDs they came out with recently... but then they are deliberately designed to disintigrate.)
Magnetic mediums still seem to be good, short term solutions for storage. I just think too much stuff on magnetic storage is designed for the long term.
Re:Speed? (Score:1)
If I understand correctly this is one of the main motivations behind Intel's Merced...err... Itanium. It's an architecture designed to be scaled up as we can make chips with more and more transistors. It's not trivial to parallelize the instruction stream.
No problem (Score:1)
This is not a CPU! (Score:1)
Researchers say that the chip stores data in the form of tiny magnetic fields
This sounds like it's merely the latest development in magnetic core memory. I had heard people were working on core memory on a chip a couple of years ago (the other Slashdot Effect: old stories presented as today's news?), so the idea itself is nothing new.
And nowhere in the article does it mention anything about the speed of the device. While I'm sure it's faster than a 15k RPM hard drive, I doubt it's fast enough to replace SRAM cache memory, or even PC100 SDRAM. Magnetic storage on a chip would be good competition for flash memory, though. Just imagine a SCSI device containing a board full of these chips.
Re:Did we learn nothing from Y2K? (Score:1)
I'm assuming by reverse you mean the direction of the magnetic field. I'm not familiar with with this, but why would it matter? Why would it be different than say, rotating your new magnetic chip 180 degrees?
Re:Bubble memory is back? (Score:2)
This has little to do with bubble memory. Bubble memeory uses a single layer of low-magnetization ferrimagnet which can hold stable "bubble" magnetic domains. Those domains were then moved around to perform read and write operations. The motion had to be slow enough to maintain the domain. Ferromagnetic materials with high magnetization can be patterned into individual section (each a bit-memory unit) and can be switched at very high speeds (less than 1 ns). They may soon compete with DRAMs, esp. when the non-volatility is a plus.
What would happen to Transmeta? (Score:1)
Re:Speed? (Score:1)
--
You're nuts (Score:2)
I think you're just trying to be a troll. I don't know why, but you can just be a troll elsewhere.
Re:microcore - can already do it (Score:3)
That's old hat these days. The latest joy in systems research is nanokernels for OPJS.
Charming [stand.ac.uk]
Re:Slow... (Score:2)
This proposed system does not have massive magnets to move around, just their fields. Beyond that, power would be consumed for the get/set currents and the switched voltages. With electronic transistors, there has to be enough electrons (or holes) to affect the electric field enough to change whether current can flow through the device. With a magnetic transistor, you can eliminate the current required to affect the switched current flow, but you can't eliminate the switched current itself. Since the switched current is not going to be used later on to switch other currents, it can be much lower than in an entirely electronic system. A lot of the micro-fields could be fixed at the factory for core logic and programming, with changable magnetic transistors just for RAM.
Right now, current has to go through a whole bunch of cascading transistors to do computation, which is why the switching rate on current chips is in the several gigahertz's while the actual clock speed is in the sub-gigahertz range. This may not be necessary with magnetic transistors, since the voltages and currents can basically just flow from one end of the chip to the other, accomplishing computation in the process. Kinda like a quantum computer (pretty similar once you consider magnetic field effect transistors to be quantum in nature).
Sounds like a good system. I liked the 'beowulf cluster of iron filings' post, wish I had some moderator points left, that was funny...
Another OS does this already (Score:1)
EROS [eros-os.org] does this. It looks very, VERY cool. Its GPL'd so I don't think anyone here will mind. Has anyone here tried it?
Speed? (Score:1)
Let me dream, 5GB mp3 player that runs off a single AA battery for 2 or 3 months.
Who are you? (Score:1)
Re:You're nuts (Score:1)
For an example at the other end of the scale, First and second generation MRI machines used large iron domes erected around the magnets. These domes would perturb and draw in the fieldlines emanating from the mri's main magnet, because iron is much more ferromagnetic than the surrounding air. It didnt shield stuff outside the dome, though it did reduce how far the field emanated from the mri magnet by essentially reshaping its fieldlines.
The amount of iron or similar material was of course proportional to both the strenght of the field emnanated by the mri, and to the amount and rapidity with which you wanted the fieldstrength to decrease past the dome (typically, the goal was 5 gauss field strength at any point where patients or the general public might be able to access, as this is the federal safety limit for pacemakers and other metallic implants).
Putting this much iron near the mri has the unfortunate side-effect of screwing up the mri magnet's magnetic field, so small bits of iron would be strategically spaced to offset this distortion (shimming).
New MRI machines usually are actively shielded. The current-carrying superconductive coils that make up the MRI Magnet (3 sets, one for each axis) essentially have a complimentary set that generate fields act in unison but opposite to the main coils, thus "cancelling out" the fields generated outside the bore of the magnet. same issues apply with one field screwing up the other, and lots of calculations and adjustments have to be made to fix things. Active shielding greatly increases the rate at which the field weakens outside the magnet, to the point where now it is easy to cram even a 1.5 tesla magnet in a truck or even motorhome-sized vehicle and drive it down the highway while the field is up.
The point of this long-winded post is that even though we are talking about much smaller field strengths, I suspect that the shielding issues will be more than trivial.
Also, I wonder how sensitive these things will be to EMPs? (Just saw "Until the End Of the World" again the other night.)
Re:Magnetic Microchips... uses thereof... (Score:1)
jeez, this was in scientific american last year! (Score:3)
Core Memory 101 (Score:3)
Core memory consisted of a number of ferrite cores strung at the intersections of wires arranged in a grid. The cores were like, little rings of material with magnetic properties.
To set a core to a 1 or a zero, half of the current needed would be sent down the "X" wire, and half would be sent down the "Y" wire - where the wires crossed (and where the core was), the magnetic polarity of the core would be set, because the full current necessary to set the polarity would then be present at the junction. If the polarity needed to be reversed, the voltage would be inverted on the wires to perform this. The polarity of the core determined whether the bit (which a core represented - 1 bit of information) was a 1 or a 0 in value.
Reading a core worked similar to writing the core, except in this case, a third wire was used. This wire was weaved through the cores in a diagonal fashion, started at one corner, and worked back ad forth through the cores to the opposite corner. The reason the wire was put on a diagonal, was to minimize the signal picked up - if it was on the same path as the X or Y wires, you couldn't use this wire to pick up the signal, because the signal would be that of the current used to flip the polarity...
Anyhow, this wire was called the "sense" wire. To see what a core's value was, the core was written to. If there existed a value in the core (the core was saturated and magnetized to some polarity), and the polarity of the written value was the same, nothing would appear on the sense wire, and so the data had the same value as what was being written. If the polarity of the written value was different, then the act of setting the value would cause a change in voltage to be picked up in the sense wire, in effect signaling that the value was opposite that of what was being written. Here is where a problem came in...
When reading a value, the value in the core is written to, and the writing to that core could cause the core to change value! This reading process was hence known as a destructive read, since the data could be changed. So, after a read, the data had to be re-written to the same core, so that it wouldn't change.
A fourth wire is also found in core memory - I can't remember what this wire wass called or what it was used for (was it a "gate" wire?) - I think it came later in core memory development, when they started making extremely tiny core systems (some of which can still be found on Ebay - man, these things are small).
BTW - I am not old enough either to "remember" core memory - I just have read enough about it, and have some really old computer textbooks and history books that explain all the concepts really well. I have been thinking about building my own small core memory system, accessing it through the parallel port or an ISA slot. I bought a whole mess of small 3-5 millimeter ferrite cores. Not small like the advanced systems were, but they don't need to be - since I will be hand threading these...
speed, mechanism, etc. (Score:5)
the BBC article is typically crap. what happens is someone from cambridge or oxford needs PR so they call up the press and tell them how many transistors they can squeeze onto the head of a pin. in the end, there's really no science in the article and, for those astute readers out there, in this particular article they don't make much mention of how these things work, what material they are using, what temperature they've demonstrated these things at, etc.
Typically, these estimates on transistor density are made when the lab produces a prototype with the active elements within a certain area. by no means does this mean that they've constructed a 5.5 billion density device that works.
they don't tell you what the mechanism is-- tunneling magnetoresistance (TMR) or spin-diffusion/accumulation ('Johnson spin transistors'), however the switching speeds are estimated to be much faster than conventional semiconductor devices (there's some argument for this in IEEE spectrum from about 5 yrs back that i can't remember).
reliability?
they have omitted mention of the gate mechanism here. how do they plan on switching these things individually? telepathy? if they are using EM fields generated by wires, then there is the inevitable heating to deal with. what material are they using? what's the curie temperature? how hot do they expect these things to get? hey wait! there's no size bar on that pretty picture of the magnets!?
blah blah blah.
BAD JOURNALISM from the BBC.
i may be a jerk about this, but i think everyone's getting a bit caught up in the hype without enough data and it's irritating as a scientist.
Re:Bulky Batteries? (Score:2)
Re:This is not a CPU! (Score:1)
Re:MagRAM? (Score:2)
conveniently it's call MRAM.
They are really sloooooooow! (Score:1)
Re:Solar? (Score:1)
Re:Say good night AC. All ACs to be banned soon. (Score:1)
Re:microcore! (Score:1)
EETimes story (Score:2)
Re:Magnetic transistors? (Score:3)
With bipolar transistor, a small amount of base current controls a larger amount or collector current. If you operate it in the linear region, you've made an amplifier. If you saturate it, you've made a switch. It's a current controlled current source.
An FET is a voltage controlled current source. A small change in gate to source voltage brings about a relatively large change in drain current. FETs can also be operated in a linear or "constant current" region. So you can make amplifiers or switches from them too.
Vacuum Tubes work similarly to FETs except that a "1" is damned big, say 100-400 Vdc! Instead of the gate and source you have a grid and cathode; instead of the drain, tubes have a plate.
... which brings me to magnetic amplifiers. The germans used these in the electrical controls of their U boats. They were totally sealed because they had no parts which would fail. They were extremely rugged, never going into microphonics like tubes would when some destroyers started pounding the sub with depth charges. Magnetic Amplifiers are made with toroidal square loop cores. A small current through a control winding established the volt-seconds of reset to the core. By varying this, much larger electrical signals can be regulated. If different core materials are substituted, it is possible to store the state of the core flux. Then you have core memory.
What these englishmen have figured out is how to microminiturize core memory without having to wind cores, etc.. Schweet!
Magnetic personality (Score:2)
Sorry, had to say it
-Joe
Holding the magnetics... (Score:2)
Esperandi
Always-ON (Score:1)
anyone for a game of magnetic life? (Score:1)
acutally have a real magnetic field effect
when storing the data from a game of life.
What heppens when the dots get realy close to each other? Do they spawn new magnegtic dots or die
off after a few generations?
Might make for an amusing project.
Whatcha see on the screen....
805,000 atoms (Score:1)
No, it was less that that (Score:1)
Moderators, take note:
1)Read the moderation guidelines before moderating anything
Re:They are really sloooooooow! (Score:2)
Re:Solar? (Score:1)
Slashdot is full of impostors (Score:1)
Re:speed, mechanism, etc. (Score:3)
Re:5,500 million transistors (Score:1)
Your budget integrated system could have everything integrated into that chip. Video, sound, I/O. Is it possible to make a chip that you could just plug your keyboard, mouse, monitor into (not the chip literally, but so there is no external logic)?
You need a certain minimal amount of external circuitry because of the various voltage requirements of external devices. Also, you still need to use an external general-purpose system bus unless you want insane numbers of pins and rapid obselecence (as in, your CPU doesn't won't connect to these VR goggles, so you need to buy a new CPU).
Re:805,000 atoms (Score:2)
It can be done... (Score:2)
Re:even better! (Score:1)
Re:Bubble memory is back? (Score:1)
I guess it was better than core...
I'll be a doctor someday (Score:1)
Still, I thought carbon's chemical shift was 20 times that of hydrogen.. I'd better go study now, thank you.
Re:Bulky Batteries? (Score:2)
Untrue. Laptop LCDs have millions of transistors (800x600 is 1.4mil) and a hefty backlight to keep things visually pleasing. Your watch/cellphone/pager LCD runs FAR less power, on the order of milliwatts to even microwatts, and the elements are often multiplexed.
Re:They are really sloooooooow! (Score:1)
The devices described in the Spectrum use a new phenomenon called giant-magnetoresistance. Those devices can be switched orders of magnitude faster than silicon devices.
They are totally different beasts.
Security? (Score:2)
Re:5,500 million transistors (Score:2)
By the way, I believe a NAND gate needs 4 fets by the way - 2 n's and 2 p's.
and 5500x10^6 is very mind boggling.
Re:5,500 million transistors (Score:1)
Thank You. (Score:1)
Re:You're nuts (Score:1)
isn't the polarity of the earths magntic field dependent on the direction of spin? that's my understanding of it, anyway. in that case, a reversal of the polarity of the earth would require a reversal of the spin, and i'm sure we would all notice if the earth suddenly came to a stop and started spinning in the opposite direction. it would probably make the chip stop working, if only as a side effect of mass destruction.
aside from that, what about the magnetic field produced by such a chip? would having so many constantly changing magnetic fields so close to each other produce any interesting effects?
Re:Another OS does this already (Score:1)
Windows 2000 does it too... but it's not GPL'd, and it's from Microsoft, so maybe a lot of y'all do mind :)
In this house we OBEY the laws of thermodynamics! (Score:1)
I sincerely hope it is the reporter and not the scientist/engineers that think this. In order to perform computation, energy is consumed. Thermodynamics says so. If you do not use up energy, you are not computing. Period.
Perhaps the reporter meant to say that it consumes a lot less energy during operation.
Free energy machine flames to /dev/null [slashdot.org].
Re:even better! (Score:1)
Magnetic Fields (Score:1)
That aside, I wonder how much energy it takes to switch these magnetic fields. And since magnetic fields I believe are infinitely reaching, how interference will affect the miniturisation of this stuff?
Also, will pregnant women not be allowed to use computers because of dangerous magnetic fields?!?
Anyway, it seems that this will be released in "several years", just about the same time as quantum computers and holographic storage!! What fun we'll have then! And we'll prolly have the tech to graft all this crap into our bodies, for those of us that like to be cut up.
Re:And you were worried about your floppy drives.. (Score:1)
Re:Body Power? (Score:1)
So if one could construct a wearable computer using this technology, you'd probably only need power for the screen and mechanical devices like CD-ROM.
If they can get this technology to read *and* write, then harddisks (which would be a bad name for them
Correct: 123,000 (Score:1)
I still think that's pretty good. Using only five-digit numbers of atoms for anything is pretty small. Interesting possibilities for future devices.
Not for military use (Score:1)
Re:Real old tech (Score:1)
This is one of the limitations of hard drive capacity, one can only fit so many bits onto the platter before the domains get too small. You can try more or larger platters but that is basicaly just adding another HD. One solution is to use a substance with greater stability at low tempratures and then heat the sector with a tight laser when you want to write.
Re:Hehehe something to hold us over (Score:1)
Re:Magnetic transistors? (Score:1)
Re:microcore! (Score:1)
So when your system freezes and won't respond to anything else you pull the plug, start again and there you are, a frozen system again. Still, if they can get past that problem, it sounds promising
Looked it up... (Score:1)
It was called the inhibit wire - and was used in a core memory "cube". This wire was threaded through the cores in such a way, on each plane in a cube, so as to "inhibit" the writing of bits on cores on certain planes of the cube. If you think of the planes as being analogous to the planar structure of, say, the VGA Mode X (all you graphics coders know what I mean), each plane is a bit array, and a word is stored via multiple planes (each plane is a bit plane). Due to the way the planes were wired (in order to make reading/writing quicker, from what I could gather), all planes were read/written at once. When writing to a plane, you needed to inhibit the current to certain cores in a word, to write a 0 in that bit position within the word. You would do that with the inhibit wire, which basically carried a negative voltage of half the current to inhibit the writing at the cores being written.
One final interesting note - something most of you may only read here. In this book, which I referenced for the info on the inhibit wire (the book was a textbook called "Computer Principles"), a mention was made about a different type of core memory - in which the cores were not individual, but was continuous. To put it simply, the device was made up of a flat plate of the ferrous material, with the wires threaded through holes drilled in the material, rather than through individual cores. It was found that the material stored the magnetic charges in the regions around the wires, and that these regions wouldn't interract as long as they were kept sufficiently far enough apart. No new name was given for this memory - it was just another type of core memory.
One more note - many of you have probably heard of mecury delay lines - but have any of you heard of nickel delay lines? Apparently, nickel (and some other metals) deform lengthwise when subjected to an electrical current. This property, with proper sensors and amplifiers, allowed early computers to have a cheaper (and less poisonous) alternative to mercury delay line storage systems. I once saw this type of system in a TI adding machine from the sixties that I had taken apart when I was younger, though at the time I didn't know what it was (this thing was completely transistor logic based - with the exception of the display - which were vacumn tube pixie lamps!)...