Slashdot Mirror
Disk Drives Face Challenge From Chips
WSJdpatton writes "Researchers are reporting significant progress in perfecting a different way to store data in semiconductors, which could replace one widely used type of memory chip and possibly become a credible competitor to disk drives. The researchers, in a paper being delivered at a technical conference in San Francisco, say they used a novel combination of materials to create prototype phase-change components that are more than 500 times as fast as flash chips, while requiring less than half of the electrical power to record data."
What is the storage density, and will it still be feasible when this finally comes to market in 10 years?
...... Should start talking about these chips being in iPods in 5-4-3-2.....
This is my opinion. To make sure you don't steal it, it's covered by the DMCA.
Making you think you're crazy is a billion dollar industry.
Not! I tried.
Interesting read, however I don't see these things holding a useful amount of data by 2010. Even if they can get 4G capacity on these chips it still wont replace hard drives that hold terabytes of data.
Although it could make really cool applications for OS installs. Could you imagine your favorite OS installed on something as fast or faster then today's RAM? I don't want to think about the cost of 4G of this stuff though. *shiver*
if I were able to see further, it was because I stood on the shoulders of Giants -Newton
From the company developing it - Ovonyx:
http://www.ovonyx.com/tech_html.html
http://www.ovonyx.com/ovonyxtech.html
Wincopy
...a mad scientist will announce to his friends and family later today that he has come up with an idea for storing 500 times more data than a DVD on a single Cheerio.
Today the bottleneck of the whole system lies in the hard drive. This is the only mechanical part (fans excluded) of a computer. It's about time to find a solution for large storage that doesn't depends on an arm swinging and moving back and forward through a fragmented file system....
It's time to realise that Abble's products are the biggest abomination these days. Just say NO to the dumb iAbble way!!
Right now are the low write count (worked upon, and mostly solved), sequential speed (again, worked upon and mostly solved) and especially the cost per gigabyte.
Will the new format allow lower costs per gigabyte compared to the current hard drive cost? Will the cost be lower than the projected cost of flash memory in the 10 years time frame? How will the cost of magnetic media storage (HDD) change in the following years?
when we all have 16 GIGS of ram and all running our OS straight from ramdrives, we will look back and laugh.
Remeber the compressor that would re-compress its output, so it would ultimately shrink to
It might be cheap:
>OUM requires fewer steps in an IC manufacturing process resulting in reduced cycle times, fewer defects, and greater manufacturing flexibility.
>a process that deviates little from a basic CMOS logic flow.
I get nervous about people who make claims like
>the OUM memory state can be written more than 10 trillion times
unless they've tested it to a trillion cycles, which is just possible.
Anyone else nervous that they didn't say anything like "write time N nanoseconds"?
Hard Disk Drives now are about $0.50 a Gigabyte. Flash is now about $25.00 a Gigabyte. 3 1/2" Floppy disks about $250.00 per Gigabyte. So it is natural for the Flash Memory cards to replace the floppies as they did. Better speed and better cost/Gigabyte. But right now Hard Drive technology is really cheap. If this new design can match prices/gigabyte of a hard drive then the Disk Drives will need a real challenge. Otherwise This new technology may only be a threat to Flash, or used with drives in hybrid mode for faster disk access. But not until then.
Price is a major driving force in memory.
CPU Registers are the fastest but most expensive (very small amount is used)
Cache is the next fastest and the second most expensive. (4 Megs or so)
Then comes normal RAM Memory Still slower then Cache and cheaper normally systems now have about a Gig or 2 of that.
If price wasn't a case Computers wouldn't have much RAM but all Cache, or huge amount of registers. But in real life price is the final decision.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
I can see as this memory becomes faster, cheaper and more reliable to replace system memory, too. I can even see the stuff become so cheap that backing all the info will become cost prohibitive, something like how tape backup systems cost way more today than a 2nd hard drive, but an order of magnatude higher.
The irony is that this would explain why in the future (à-la-Star-Trek), backups of the computer's memory doesn't exist and cause improbable storylines for us system admins.
Obama's legacy: (N)othing (S)ecure (A)nywhere and (T)error (S)imulation (A)dministration
One of the big failings of flash memory is the limited number of rewrite cycles. HDs can be rewritten many times without going bad. How many rewrite cycles will this have?
Best Slashdot Co
As I recalled, flash is only good for 100 thousand writes. Given Windoze freqently write to the NTFS volumes even when the system is idle, it won't take long for the chip drive to become useless.
The quicker we can get away from a spinning rust platter read by magnets, the better. Less moving parts = more reliable (in general.)
Can't we just skip ahead to the transparent crystals that glow in various colors and store almost limitless data? We all know that's where this is heading.
Maybe we need to perfect holographic 3D displays first?
The results are presented at the IEDM conference, and it seems that there's no published article on this yet. From this page I get:
Ultra-Thin Phase-Change Bridge Memory Device Using GeSb
Y.C. Chen, C.T. Rettner***, S. Raoux***, G.W. Burr***, S.H. Chen, R.M. Shelby***, M. Salinga***, W.P. Risk***, T.D. Happ*, G.M. McClelland***, M. Breitwisch^, A. Schrott^, J.B. Philipp*, M.H. Lee, R. Cheek^, T. Nirschl**, M. Lamorey^^, C. F. Chen, E. Joseph^, S. Zaidi*, B. Yee^, H. L. Lung, R. Bergmann*, and C. Lam^, Macronix International Co. Ltd., *Qimonda, **Infineon Technologies, ***IBM Almaden Research Center, ^IBM Watson Research Center, ^^IBM Essex Junction, San Jose, CA
An ultra-thin phase-change bridge (PCB) memory cell, implemented with doped GeSb, is shown with 100microAmp RESET current. The device concept provides for simplified scaling to small cross-sectional area (60nm squared) through ultra-thin (3nm) films; the doped GeSb phase-change material offers the potential for both fast crystallization and good data retention.
Any sufficiently advanced libertarian utopia is indistinguishable from government.
Still won't be able to compete with the sheer density of colored symbols on A4 paper.
The only real competitive advantage discs (optical or magnetic) have is cost.
Slow, low density optical discs are good for offline storage, up to 4.7GB at about $0.042:GB. Plus about $1000 for a 400-disc changer jukebox makes about $0.60:GB across all jukebox loads, theoretically also automatable across many loads, for "nearline" storage.
Fast, high denisty magnetic discs are good for online storage, the kind we use as "permanent" without worrying about dealing with them directly (until they fail). They cost about $0.23:GB.
Flash currently costs about $14.00:GB. Obviously archive or real longterm storage isn't threatened right now, except in mobile devices (not just portables with biggish/hottish HDs).
But really mobile devices will have just storage of secrets (keys), pointers (URLs), wireless network interfaces (or HW jacks for the paranoid), and auth UIs (like thumbprint or other biometrics, and maybe still passwords). Because generic computing/comm devices will be everywhere, immersed in wireless networks. Discs have to rotate inside something, but why carry that everywhere, especially when it's fragile? And large capacity is unnecessary in personal tokens, with other tech distributed around the Net.
So while solid state storage is becoming cheaper, the infrastructure that makes it really cheap and easy is growing even faster. By the time a personal token costs $1:GB, it will include wireless/auth interfaces to a ubiquitous wireless Net. And maybe all those spinning discs will go the way of tape: specialized apps that require extreme density, and specialists to operate them.
--
make install -not war
If these phase change chips are non-volatile and cheap, they can replace the DVD/CD just as easily as they replace the hard disk. The only purpose of anything DVD-like would be for mass distribution of software and content, and if I could eliminate that mechanical drive in favor of simple internet downloads to a phase change stick like current flash sticks but faster and cheaper, I'd be happy.
Infuriate left and right
My first Vax, 22 years ago, had 1GB of disk, in the form of four washing-machine-sized drives which used removable 250MB disk packs. The drives cost about $120K total, and the packs were about $1000 each. There isn't really an exact comparison to that combination; you could either look at DVD-RW ($40 for the drive, $0.50 for the disks, so 8-12000x the price/capacity), or amortize the drive across some number of packs to compare to fixed disks (e.g. 10 packs per drive would be $160K for 10GB, though I think we only bought about 3 packs per drive over before that machine was obsolete), or you could make some unbalanced comparison like $20 for a CF-to-USB adapter and $20/GB for Compact Flash cards, which would be a mere 200:1 on the removable media but 6000:1 for the "drive".
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
So run out, children, and buy your SD 2.0 standard devices while they're not yet obsolete. That way you can buy your camera again and again for no good reason.
Help stamp out iliturcy.
Page 35 of their downloadable pdf shows that each cell can hold multiple bits. Each cell can be set to one of ten states by multiple pulses of current, so comparisons to binary storage don't work. The manufacturing process is not complex, basic CMOS in about 20 stages, but the part of the cell that stores data is only about 20 nanometers wide. Replacement of hard drives is a very trivial application. IBM and Intel are planning to incorporate this tech inside ICs to reduce latency of fetching data. The big news is more highly integrated systems on chip. It doesn't look pie-in-the-sky, somewhere-way-down-the-road to me.
Why haven't we thought of this before ??
I'd guess that the data would be highly compressable, though: Dark, dark, dark, dark, dark, dark, dark, dark, dark, star, dark, dark, dark, dark, dark, dark, dark, star, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, dark, star, dark, dark, dark...
When our name is on the back of your car, we're behind you all the way!
...'cause where you gonna put the salsa?! Idiots.
How many libraries of congress are there in a cheerio?
"A terabyte is NOT 128 gigabyes. A terabyte is NOT 128 gigabytes...."
A terabyte is 1000 gigabytes (1024 if you're old school), not 128.
When our name is on the back of your car, we're behind you all the way!
I'm so tired of finding the word "novel" in research publications, etc. It's an overused word which contains absolutely no new information. Most style guides for scientific writing grip about it, and many research publications ask authors not to use such words in titles or abstracts, but I get the feeling the use is still increasing. It's basically scientific marketing, and it's not even clever or original. Most things which are described as novel are anyway anything but, and the word has just become weak and watered down. The sentence "used a novel combination of materials..." could have been replaced with one containing some useful information, or just done away with alltogether to improve reading efficiency.
Yeah, it's probably completely off topic, but I just had to say it.
"The typical access time for a Flash based SSD is about 35 - 100 micro-seconds, whereas that of a rotating disk is around 5,000 - 10,000 micro-seconds. That makes a Flash-based SSD approximately 100 times faster than a rotating disk. " http://www.storagesearch.com/bitmicro-art3.html
h tml
"The SSD found files more than twice as fast, and accelerated boot-up. Its cumulative speed advantage over the other two drives was an impressive 25 percent" http://www.pcworld.com/article/id,126833/article.
Solid state drives (flash drives) are not slower. If they were, hard drive manufacturers would not be using flash to make hybrid drives.
When our name is on the back of your car, we're behind you all the way!
Has your company never heard of Wake on LAN. If there is an upgrade to be pushed, just send a few packets over the network to ensure all the machines are up and running, then push the upgrade.
When our name is on the back of your car, we're behind you all the way!
While the actual flash technology might be capable of that kind of speed, the entire stack isn't. Compare the MB/s throughput of several hard drives here with the throughput of several USB flash drives here (both benchmarks done with SiSoft's Sandra).
Bottom line: The USB drives are topping out at an average of 8 MB/s, the hard drives are in the 60 MB/s range. That alone puts hard drives an average of 7.5 times faster.
Flash drives have great single block seek times because they don't have to move a head, but most benchmarks show that their ability to move large quantities of data quickly sucks.
Have you ever considered piracy? You'd make a wonderful Dread Pirate Roberts.
A really fast 40G drive would be great to use as the filesystem root, plus swap space; your porn and other documents could all be kept on another (large, slow) drive. I've thought a lot about doing something like this right now using SCSI disks.
/usr (or partitioning it in half and mounting at /usr and /home, or whatever). While this is possible to do on MacOS, and I assume it's possible on Windows, it's obnoxious because those OSes are created with a single-drive-system approach in mind.
One of the biggest advantages of Linux that you never really hear about is the ease with which you can create a system that spans multiple disks, keeping frequently used (OS, libraries, swap) items on a fast drive and application data and documents on another one. It's a trivial matter of mounting the big drive at
I think what we're going to see in the future is more specialization in terms of which technologies are picked for different uses. Right now, hard drives are "good enough" for lots of types of storage -- from occasionally-accessed data that could probably be moved to offline or nearline storage, to VM temp/swap files. As people and developers start demanding more performance out of systems, this compromise solution may start to look worse and worse. The upcoming hybrid (memory+platter) drives are only the beginning. I suspect that we're going to have to rethink the one-size-fits-all approach to storage, and thus make the speed/cost trade-offs independently for each type of data we need to store.
"Ladies and gentlemen, my killbot features Lotus Notes and a machine gun. It is the finest available."
FAT is both unreliable and slow. Most **real** FFS are log structured which gives them better performance and robustness than FAT.
Engineering is the art of compromise.
If it's 500X faster at .5 power, does that mean it needs 250X the power of flash for a much shorter duration?
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Since I worked in the Disk Drive Industry I was putting together a more meaningful and 'interesting' post but then my ADD kicked on and you see what you got. FWIW Hard Drives are always going to be a couple of generations ahead of Solid State Storage. HD space is down to approx. 24 cents a GB. I don't see Solid State at that level for quite awhile. Oh, gotta go, shiny object.
Beyond the obvious potential benefits of tossing out a mechanical device from a computer (power, speed, reliability, weight)... if all of our storage is basically running at the same fast speed, would there be any need for VM anymore? Perhaps we could just fragment the memory system into parts which could be used by the operating system, vs. those for user space, with appropriate protection mechanisms placed by each process on it's allocated (but physically addressed) space. There are plenty of 'advantages' to using VM, but each layer of goo we put on top of the underlying hardware causes a performance penalty (both because of the extra layer of indirection, and the abstractions taught in school make it almost a sure thing that kids today will write really inefficient software). We'd need to keep processes which weren't supposed to write all over each other from doing so with a new form of protection... Maybe it's just that I miss poke and peek, but I think I'd be happy to ditch VM as we know it along with the spinning platter.
No more printed-paper storage?? But if I've just heard it is the future!
Ferroelectric phase change densities of .2 to .5 Petabits = 200 to 500 Terabits sq. in. / 40 Petabits = 40,000 to 100 Petabits = 100,000 Terabits cu.cm. or 200,000 to 500,000 Gigabits sq.in. / 40,000,000 to 100,000,000 Gigabits cu.cm. with symmetrical read / write times of 160 picoseconds for 100 year non-volatile storage having infinite rewrites.
.4 nanometer cell size.
Normally the 1.3 to 5 nanometer molecule can switch at 160 picoseconds while maintaining non-destructive readout of ferroelectric bistable properties at a 5 nanometer cell size.
This is not the end by any means as Tohoku University says their target is 4 Petabits a sq. in or 375,000 Terabits cu. cm. using a
Ferromagnetic - Magnetic disk drive have the next highest density at 50 Terabits sq. cm., The Super Paramagnetic Limit. This density is a particle size prediction ( 10 nm ) where the real operational density is more around 1 Terabits sq. cm. as 50 nanoparticles are needed per 1 data bit cell ( ~ 500 nm ). Perpendicular orientation recording will only delay this dead end a few years at most.
Ferroelectrics will not use complex wiring schemes that have impedance, heat, and reliability problems but instead will exceed any storage devices proposed and would enable a storage density of more than 100,000 terabits per cubic centimeter. A ferroelectric storage drive device the size of an iPod nano or 3.5 inch drive could hold enough MP3 music to play for 300,000 years without repeating a song or enough DVD quality video to play movies for 10,000 years without repetition.
that the Journal of the American Chemical Society (and subsequently all of its daughter journals) banned "novel" from titles a couple of years ago.
This one "http://www.thinfilm.se/default.aspx" has been around much longer and has much better graphics. Actually, the contents have not changed since I first looked at it...so many years ago that I cannot even find the article from the register anymore. It's too long ago. Also, Intel used to invest in this technology. Just to give you an idea on how much vaporware this could be.
How would this work? Would the user take the bus to the video store to load up a phase change stick with a movie? Or would the user pay the cost of several DVDs per month to set up high-speed Internet access? (DVD is 10 Mbps; low-end broadband is only 0.8 Mbps.)
If one cycle takes a microsecond (10^6 Hz), then a teracycle (10^12 cycles) test takes a million seconds (10^6 s), or less than twelve days.
In legalese, a process or composition of matter is called "novel" iff it is not taught by the prior art. Researchers who produce both papers and patent applications likely use the word out of habit.
Other than lower power, possibly smaller size, I see one of the biggest advantages here is the noise reduction. Hard drives, in my opinion, tend to be the noisiest part of any PC, and the reason why it's annoying having one in your living room as a media PC, or one in your bedroom (come on geeks, you know you all have one in your bedroom).
Next to the hard drive, fans tend to be the other noisy component, but there are lots of existing solutions to that problem (higher quality, larger, quieter fans, other cooling methods, etc.)
Love many, trust a few, do harm to none.
It should be possible to make terabytes of it, on one chip, now, by using this stacking technique:
http://memory.oyhus.no/
Kim0