Slashdot Mirror
Nanomagnets for Hard Drives
Single GNU Theory writes "Scientists at Cornell have developed magnets so tiny they could be used to create standard form-factor hard drives with terabyte capacities. All they need is a way to read and write them fast. Check it out here.
Hmmm... That'd require 250 Amiga Fast File System partitions! "
4k is the default size, but you can make a fat32 cluster any size you like.
That's right boys and girls, Microsoft has innovated another product, MS Diskfiller. Did you know that most of your TB hard drive is being wasted? That's right, wasted! You paid good money for it but it's not doing its job because it is not holding a terabyte. Simply insert the MS Diskfiller DVD into your DVD-ROM drive and we'll take care of the rest. Using a patented algorithm, MS Diskfiller is able to fill up your disk faster and more innovatively than other competing products so that 100% of your drive will be utilized. Did you know that you have been using a similar technology for years? That's right! Our popular Windows brand of operating systems have intentionally been coded to use up all of your CPU! Those other operating systems have been wasting the money you spent on your CPU by letting it sit there idle, but due to the innovative programmers at Microsoft, we are able to give you your money's worth.
This innovation has been brought to you by the innovative folks at Microsoft.
Did you forget that light travels at a finite, although constant speed? I.E. it won't be zero time. Not to mention the fact that a CPU doesn't "spin" anyway, and the electricity that makes a CPU work is, of course, already moving at the speed of light.
-xnec@infernoDOTtusculumDOTedu
Fsck me, that was fast...
Um.. I hope someone can e-mail the author and tell him that it's not an alien / ST user attack...
for the distribution :
www.aminet.org
www.uk.aminet.org
sunsite.doc.ic.ac.uk + click on Aminet - amiga related files...
Then aminet search SFS for the distribution.
FAT32 is *not* a good idea. The FAT filesystem stores directory entries unsorted, making finding a file in a directory an O(N)-time search, and also stores files as linked lists (ie, to seek to position K in a file, you have to follow a chain of FAT records), making "random" file access also an O(N)-time search. These are both O(log N) operations on any other filesystem. FAT is also extremely unrobust. Metadata corruption is difficult to detect, difficult to repair, and tends to breed additional corruption within the filesystem with normal use. .. and I've been wanting to get back into linux kernel hacking (after about a three-year haitus). Hmm..
On the other hand, ext2fs is showing its age. It's high time we incorporated some modern filesystem technology into Linux, and started taking advantage of today's high-capacity drives to store additional error-correction metadata. There is no good reason metadata corruptions could not be detected and corrected "on the fly", transparently to the user. There are some other improvements I've been wanting in my desktop's filesystem, too
-- Guges --
Comments in the mailing list suggest that future versions will provide on-the-fly defragging.
Yeah, PFS2 is actually very good too - but the gist of my posting was that since SFS is free, some of it might be useful from a linux perspective too - and how it works is detailed openly, so even if SFS itself is never used wholly, some of its design concepts might be very useful in a non-amiga environment, too.
CPU's will spin at the speed of light which will make the compute time for anything 0.
Seriously, ext2 is outdated in the face of mere current technology. > 8GB hard drives are commonplace now. Maybe we should jump on something like UDF (the filesystem on DVDs) which can support much larger storage. Heck, even FAT32 is beating out ext2fs.
I believe Apple's Sherlock does exactly what you are thinking of.
Check out www.nanomagnetics.com. Similar solution, but smells more like nanotechnology to me.
We already have "microdrives". Unfortunately they are relatively expensive ($150 for a 500MB drive). The bulk of the cost of hard drive systems isn't in the magnetic media; it's in the control logic and mechanical devices which spin the platters and move the arms oh-so-precisely. The price of a hard drive will never drop below $100 or so, no matter how small they get, because the interface, the logic, and the mechanical components set a lower boundry.
What technology like this will do is make the price-per-capacity plummet, but the price of a desktop's hard drive will always be around $200 (ie, the price will remain the same but the size of the drive will increase), unless someone comes up with something cheaper than spinning a platter and moving an arm.
-- Guges --
Well :-) I've noticed that this happens on *most* people's systems, and I silently cheer them on. My hard drive needs are much more modest (I don't play games, I don't use MS OS's, I don't tend to use/have any graphical images, and my code and data structures are fairly tight and lean), so I'm rolling in riches, space-wise. As long as everyone else keeps driving hard disks into bigger and cheaper ranges, I'll be pretty happy. The 5GB hard drive I bought more than a year ago still has almost 2GB free, and believe me I've been *very* busy writing software, generating data, and even pulling (textual) documents from ftp/web sites. My biggest problem is keeping it all organized and finding the time to sit down and read through or otherwise analyze the information I already have.
-- Guges --
network problems my end...
You have noticed this trend also. I remeber having game on audio tape when I was using a Timex Sinclair 8000. Then my first pc had 20 meg hd's, those cost $300+ a peice at the time. But on that 20 megs I could fit alot of things on it.. Now 20 megs holds about 4 mp3 files or a bunch of sciptz kidz files (which I do I have a couple of these to do security for freinds, using their files to foil them). Now I have a 1.2 gig and a 2.1 gig in my box and I find that my 1.2 is 90% full and I barely have anything on it! I think i'm going to get a 12gig drive when I build my new box in couple of months. Programmers should learn how to optimize the size of their code w/ out compromising the speed or the operation of it.
I ate my tag line.
-=Ellis (D)25=-
Well my answer would be to have a seperate proccessor just for that.. Maybe a 128 bit 300mhz risc chip.
I ate my tag line.
-=Ellis (D)25=-
Fat32 is good idea, but they need to make the chunks about 1k, instead of 4k (if I remember correctly).
I ate my tag line.
-=Ellis (D)25=-
SFS is a free implementation of a "cursed l:disk-validator"-free filesystem for the amiga. disk-validator is the equivalent of fsck, but for amiga OFS and FFS. In later amiga models, it was in rom rather than in l:
SFS uses the 64-bit Amiga New Style Device (NSD) API that was published a while after CBM went belly-up, and claims to support up to 2000 GB partitions, though limitations in the current dos.library prohibit individual files > 2GB long (note to linux-people: this is not a limitation of the filesystem itself, just the amiga's antiquated 32-bit dos.library v40.x (yes, the amiga version numbering conventions are a little different to linux))
It's available from the link below. This site also has a remarkably easy to follow description of how the filesystem actually works, including some rather good diagrams.
http://www.xs4all.nl/~hjohn/SFS/
I've been using it for some time now, on my poor lonely old 68060 amiga (that sits next to my linux-x86 box) and it's wonderfully fast compared to amiga-ffs, more space efficient, and I've deliberately punished it by turning the computer off in the middle of writes, without any loss of data coherency, and pretty much immediate boot-up afterward.
It will also support the (relatively little-used) amiga multiuser filesystem bits+pieces, too! ( thus making a linux implementation even more useful - it could be used as a native partition with no loss of functionality)
It actually seems to work better than the current ext2fs, IMHO, despite being nominally a late beta.
It would be great if some linux next-gen filesystem developers were to at least look at the site, if only to make sure SFS hasn't got anything
they haven't thought of.
Now, how about a linux implementation ???
This wouldn't help a bit. The data would just expand to fill up all the space you can get.
Side note: we really should create an altavista-like tool to index our HDs before we get truly mad because of the excess of random assorted huge junk files.
Posted by himes:
400 x 400 nm bits in a hard drive aren't a big deal. You'll be buying the same areal density from Office Max in the very near future.
When it'll be ready, the question is how much it'll cost? Some great technology never see the light because they would be too expensive to produce...
But they need to sort out the processing power.
The site has been slashdotted already.
NEver underestimate the power of Pron, MP3s and \/\/4r3z [ok and any MS/OS] to fill up a HD that large in minutes. This will however be pretty cool for video storage [assuming it's fast enough]. Maybe we'll be one step closer to *real* video on demand.
Blocklevel: Practical Information Architecture
With this technology, couldn't you also use the laser to burn a pattern of "live" and "dead" spaces into the medium, thus getting a read-only medium with even higher density? Or do you only get parallel lines of scoring?
:-)
no, that's not possible. all you can do is etch out lines, and by repeating the process rotated 90 degrees you can etch out a grid, so you create little islands of magnetic material.
Also, from what you're saying, this is a grid, not a radial pattern like what you'd need for a rotating disk medium. That might actually be better, though: what kinds of bearings could rotate a disk with only nanometers of vibration?
yup, it's a grid. but as you say, a rotating disk will undoubtedly vibrate, which probably is very bad for the head's accuracy
)O(
the Gods have a sense of humor,
Never underestimate the power of stupidity
To err is human, to moo bovine
but then you'd first have to make a flat magnetic layer and then wrap it around a drum, and I don't know how possible that is :-)
)O(
the Gods have a sense of humor,
Never underestimate the power of stupidity
To err is human, to moo bovine
well who says they're gonna put a full size hard drive in a portable MP3 player? just think how small a 5 or 10GB drive would be using that technology. tiny drives like that would revolutionize portable equipment like laptops, palmtops and MP3 players. not to mention that a drive like that would require much less power, which would also be very convenient for portable computers.
)O(
the Gods have a sense of humor,
Never underestimate the power of stupidity
To err is human, to moo bovine
You're right, of course. Oops.
Jon
All opinions expressed herein are my own, and not those of my employers, who are appalled.
I could see someone building a hard drive for a mouse- or bug-sized 'bot with this tech, but not a viral-sized one!
At one molecule per bit, it's going to take a lot of nano-legs to carry the code around!
It's kewl, though!
Jon
All opinions expressed herein are my own, and not those of my employers, who are appalled.
Not too good of an example. In terms of pure text, a book should easily fit on the floppies we have now. Illustrations would bulk it out a lot more, of course.
;)
Oh well, it's only journalism.
Jon
All opinions expressed herein are my own, and not those of my employers, who are appalled.
With this technology, couldn't you also use the laser to burn a pattern of "live" and "dead" spaces into the medium, thus getting a read-only medium with even higher density? Or do you only get parallel lines of scoring?
Also, from what you're saying, this is a grid, not a radial pattern like what you'd need for a rotating disk medium. That might actually be better, though: what kinds of bearings could rotate a disk with only nanometers of vibration?
Jon
All opinions expressed herein are my own, and not those of my employers, who are appalled.
Actually, downloading the entire internet has already been accomplished. In parallel.
If tits were wings it'd be flying around.
This is "microtechnology" or whatever you call top-down construction. Nanotechnology implies creation from the bottom-up (the atoms are combined instead of removed). Sorry, folks.
:P
If you think I'm wrong, then too bad.
Hey, now we can actually test that claim of BeOS
that they can support gargantuan file systems.
Just how many nanites can dance on the R/W-head of a hard drive? And will looking at it while it runs make it disappear here, and appear again in Alpha Centauri? And do we really have to stick a cat in the PC case to make it work? Will it work faster if travelling at the speed of light? And if it weighs as much as a duck, does that mean it is made of wood, and therefore a witch??
:)
Things are getting freaky. But, I'm looking forward to downloading the whole internet.
-- What you do today will cost you a day of your life.
Build a bigger hard drive and the world will build bigger files. Soon we'll all be complaining about how our new 1.2 TB drive is all filled up and there's nothing we want to delete.
*begin nostalgic music
I remember when I got my first 150 meg hard drive. Almost 4x larger than my 40 megger. "I'll never fill that up! " I foolishly said as I pulled it from the box, carelessly tossing aside the $500 invoice. That hard drive is still running today in that old 386... ESDI was the great new technoligy NEC had said....
*end nostalgic music
Programmers should learn how to optimize the size of their code w/ out compromising the speed or the operation of it.
Isn't that the sort of thing that the first small groups of hackers were proud to have done, when the only available space was about 1-3K? When you had to fit an entire video game in 2048 bytes, and used all sorts of interesting tricks to do it?
I remember this from a book on the subject, and a few such tricks were mentioned:
Using data as instructions (weird but useful in a fractal sort of way)
Understanding and using the full instruction set of a CPU (makes for proprietary code, but we're talking ASM at the time anyway --which is much more efficient just because of what it is)
Simple compression algorythms (using the small free memory space to store uncompressed pages of code)
Overlay files (Inefficient use of disk space, but efficient use of RAM)
and so on...
Even with enormous sloppy APIs, certain things don't ever go out of style... like well-organized program structure, efficient algorhythms, et cetera.
But you knew that.
Actually, rotating r/w heads are an excellent idea. Just pop in a square chip with a notch in the corner, and the two heads will lock in and spin like cyclones...
:)
And as discussed elsewhere, with multiple heads on each platter you have vastly-reduced seek times and potentially extended disk life (the arm doesn't vibrate so much when it's a disc).
Then again, the head doesn't have to rotate, necessarily. In fact, it might just be another plate of magnets, cone-shaped this time. Kind of like old-time core memory (mini-donut magnets wrapped in wire), we would in fact have a new kind of stable RAM chip. 2Gig flash RAM, anyone?
The article says "A single 1.44 megabyte floppy disk can hold an entire book with room to spare; a terabyte hard drive could hold a library with at least one million titles." meaning that currently, this is true. I thought the same thing at first myself, then reread it.
] D
I'm all of 26 years old and I recall playing with a PDP11 in my father's office. Ah, those were the days of platters and the clackety-clack of the punch machine. Nothing more fun that making lace cards to piss my pop off. =)
But MY first drive was a 40m'er. I donated a 100m to a local bbs and was immediately given god status. That was way back in...what? 1991? Heh.
My
Quux26
www.crashspace.net
George Lucas should be happy. Unfortunately it seems rather far off right now.
--
JCA
the best part is not that drives will now be in terra bites (excuse my sp) ... because no matter what, the bigger the hard drive the bigger the files ... microsoft alone will make the first terra bite word processor, i'm sure ... games will be bigger ... we are like gold fish ... we expand to fit the space we have ...
... storing media in archives is always the most obnoxious part for both buisness and users ... zip disks cost entirely too much to truely archive what you want to archive ... but floopies? ... if one floppy can hold more than one present zip? ... that would be cool ...
... but i'd love to carry around the works of shakespeare on a floppy ... or how to program in ___ (enter programming language here) ... then we could start making little book reading thingies ... that instead of carrying heavy books around you carry a viewer (bookman vs walkman?) and some disks ... you could carry 20 books and still have room in your back pack for jolt and some peanuts ...
..
the truely great part will be the floppy disks
granted the games still won't fit on floppies because they will have expanded
maybe i'm the only one obssesed w/ reading
if knowledge is power, the internet is god - me again
I'm not sure this nanomagnet scheme would
be feasible. When the domains get small
enough, they can flip between states by
quantum mechanical tunnelling.
A scheme that *does* look feasible is storing
of individual electrons in nanometer-scale
metal particles. This would be electrostatic
rather than magnetic storage (but would still
be on the surface of a rotating disk). The
read head would be a "single electron transistor"
with a head/disk spacing similar to current
hard drives. See the most recent Proceedings
of the IEEE for an article on this.
From my own studies into molecular systems, I have to say this will be a large step forward in developing nanotechnology. Now that is possible for us to make such large capacity hard drives, we can decrease the size of the drive and we'll have perfect data storage for nanocomputers and nano-robots. The possibility of a watch containing more data storage as my computer has now, or an internal medical nanobot with a database of biological and medical knowledge, it is all suddenly viable (assuming that other key components of molecular systems are advanced as well). All in all, not only is this a good thing for the computer world, but ultimately for any field that molecular systems will touch, which is basically any and all.
Jack
Demona's Law - "User data expands to exceed available bandwidth." ("User data" being pr0n, mp3's, vob's,
Simple, convert a nanotech assembler to read data instead of moving atoms...of course this would be a LOT simpler if assemblers existed right now. Doh. Oh well, history has shown us that we have the right ideas for technology, it's the implementation that screws us up. Best of luck to the researchers out there in developing this technology.
Demona's Law - "User data expands to exceed available bandwidth." ("User data" being pr0n, mp3's, vob's,
But the theory of computation still applies... there are some problems that will always be unsolvable via a computer, even if one can go arbitrarily far back in time arbitrarily many times. Whatever fancy computer you come up with, if it is a universal one (one that can emulate any other computer of its kind), I can just feed that same computer its own encoding, and some string, and no one (i.e. no computer) will be able to tell if that computer will ever stop running. In finite time.
Things are getting freaky. But, I'm looking forward to downloading the whole internet. :)
;)
Hey, I've done 42% already
I just found this out from a friend who's studying electrotechnics here; the University of Twente, Netherlands, is developing a similar system.
:-)
with current technology all domain grains are irregular, and the stronger ones might start dominating the weaker ones over time, corrupting the data, so you need about 100 domains per bit, making one bit about 50nm in size. the solution to that problem is to make sure all domains are equal in strength. up to here it's the same as the nanomagnet theory.
the next problem is how to get those grains on a drive, and here is where it starts differing. the nanomagnet technology takes a number of nanomagnets and lines them up. the technology developed by the UT uses LIL, Laser Interference Lithography. the surplus magnetic material is etched away using two interfering laser beams, so in the first go you create lines of magnetic material. then you turn it by 90 degrees and repeat the procedure to create small islands of magnetic material, at perfectly equal distances. currently they're able to create a periodicity of 200nm, but with a new laser they could decrease that to 160nm, and with a new technique that is still in development they could decrease it even to 80nm. yes, 80nm per bit.
they've also designed a head with a one atom wide tip to read the data... the only real problem still is how to position that head accurately enough and quickly enough to allow for reliable, efficient and quick reading and writing.
so basically the University of Twente is already ahaid of Cornell University
)O(
the Gods have a sense of humor,
Never underestimate the power of stupidity
To err is human, to moo bovine
this technology won't just enable you to create huge storage capacity in a standard size hard drive, but also teeny tiny hard drives with standard storage capacity... how big would a nano-magnet drive with 10GB capacity be? 0.5x1x2" maybe? imagine what that would do for portables and wearables. and a tiny drive like that would surely also take less power, yet another convenience...
)O(
the Gods have a sense of humor,
Never underestimate the power of stupidity
To err is human, to moo bovine
A terrabyte for portable MP3 players? Does anyone have that much music? Doing some quick math and assuming 1 minute = 1 meg (although you'd probably use something better if you had this capacity).
:-)
You'd be able to store 1048576 minutes of music, or 17476.26666667 hours, which is 728.1777777778 days or (assuming 365 days per year), 1.99500761035 years of music on a single disk.
I don't think that's the most pratical application of this technology
With this areal density, your wouldn't go for today's form factors. You could make a thumbnail-sized drive that would hold a few gigs, assuming that you could shrink the electronics, rotor, and servo down to similar scales.
One pressing question is whether today's rotary voice coil head actuators are accurate enough to position the head within +- 20 nm or so. Probably not.
If you check out the original site at Cornell you'll find this link where they talk about growing nerve cells on silicon. William Gibson's predictions coming true already....
--a proud Cornell class of 1970 alumnus
Thank God I'm an atheist!
Still reason to be optimistic, however.
They can't yet lay down a layer of these things into the track/sheets needed for a disk plater, for one thing. They had to do so individually, best I could tell, when what they want is to, perhaps, apply some sort of magneto-electric field, spray a mist of these magets embedded in a gel, and as the magnets settle towards the surface of the platter they'd align and space themsevles according to the field, and then as the gel/platter is heated the fields become erased and the surface solid and fixed for future use.
Then the problem of actually reading/writing to the disk.
The suggestion of a massively parallel fixed read arm would still have the problem of many 'wires' of the same size as the magnets, which might not be feasible.
If a motorized arm is used, we'd need something more precise and accurate than currently possible. Perhaps piezo-electric seek heads, grown within specifications, used to move the arm back and forth. As opposed to coils or something.
If I make no sense, please ignore, but I think those are some of the biggest hurdles to deal with.
AS
-AS
*Pikachu*
Presumably, if the heads are small there's less mass to crash into the magnetic surface (although less inertia to resist shock in the first place, of course) - which implies better shock resistance since the drive will get gouged less when you drop it.
Not that I'm in the habit of dropping my hard drives, of course, but it makes high-capacity MP3 players cheaper.
Dunx
Converting caffeine into code since 1982
Haven't used SFS (though I've heard a lot of good things about it); however, I do use PFS2 on my Amiga's hard drive and Zip disks. It's a similar idea to SFS; a file system that is organized in a less stupid way than FFS, and promises data coherency no matter what horrible things you do to your machine. It works perfectly; when the Amiga crashes in the middle of a write (which happens a lot to me, since there's no memory protection and I do a lot of silly things), it boots straight back up again. No corrupted filesystems, no waits while the filesystem is revalidated.
It's waaaaaaaay faster than FFS too, and has a hidden directory in the root of each filesystem which stores copies of the last thirty deleted files, so people who rm first and think later (like me) don't have to spend ages trawling through backups ;-)
I strongly recommend Amigans who still use FFS to check it out. UK Amigans can buy it from Ramjam Consultants.
How are we supposed to replace our magnetic media with something more esoteric when they keep making magnetic media more esoteric? Next I suppose they'll have little nanomachines pumping pedals to spin the drives...or they'll rotate the heads instead of the platters.
Yes it is possible with a cartesian grid. Put the grid on a magnetic drum, not a magnetic disk. Magnetic drums are particularly fast with head-per-track designs.
It might be zero time. The "teleportation of light" experiments in the past few years are altering a characteristic of light at a sensor. The researchers are considering using this for communication and computing. For communication, an entangled light beam could be sent between two places and alterations teleported between them...if entanglement works in zero time, bits could be sent across the country in zero time although the entangled light already made the trip at the speed of light.
Why have a rotary voice coil. One of the limiting factors in the drive is positioning the coil accurately. Just put an array of heads across the disk on a static arm -- one head for each track. You get an average seek time of only 1/2 rotation and no alignment issues. A lot of complex circuitry is eliminated, further shrinking the drive. Futhermore, you could get 1024 tracks read in parallel -- transfer rates would skyrocket.
I'm sorry I have to stop now, I'm drooling on my keyboard.
Aah, change is good. -- Rafiki
Yeah, but it ain't easy. -- Simba