backtop


Print 41 comment(s) - last by mathew7.. on Apr 14 at 3:46 AM


IBM's new racetrack storage takes advantage of advanced quantum mechanics to move tiny magnetic domain walls on which information is stored.  (Source: IBM)
IBM bets big on spintronics, new type of memory which it says will deliver terabyte MP3 players

With computer hardware ever advancing and improving steadily, the question in the hardware development world is not so much "if" but "how?"  IBM believes that it has developed the storage solution for next generation of consumer electronics and computers.  IBM's new storage technology is known as spintronics, and it is betting that it will replace both the solid state drives (flash) and traditional magnetic drives, by offering higher densities at a lower price.

The basic premise of IBM's technology is storing the data on a wire track.  IBM calls the memory "Race Track" memory due to the fact that the data "races" around the memory's track design.  It was developed by IBM Fellow Stuart Parkin and his colleagues, whom are firm believers that the new technology will make older storage methods obsolete.  According to Parkin the benefits are across the board-- much improved read/write times, higher capacities, improved stability, improved durability all at a lower cost than today's memory.

Solid state electronic devices based on the new memory could theoretically store 500,000 songs or around 3500 movies on a standard MP3 hard drive, and be accessed at lower power consumption.  The data would be safe, according to IBM from degradation for decades.  Further the power consumption would be so low the battery could run for weeks on a single charge.

While Parkin's enthusiasm is certainly infectious and the idea remains intriguing, it turns out the technology relies on some exotic science that prevents it from being commercially realized today.  The new memory utilizes a field of quantum computing known as spintronics, which involves reading and writing data by altering an electron's spin.  Currently no known commercial ventures using such technology exist, so its commercial implementation remains far off.

Undeterred, Parkin states, "It has been an exciting adventure to have been involved with research into metal spintronics since its inception almost 20 years ago with our work on spin-valve structures.  The combination of extraordinarily interesting physics and spintronic materials engineering, one atomic layer at a time, continues to be highly challenging and very rewarding. The promise of racetrack memory - for example, the ability to carry massive amounts of information in your pocket - could unleash creativity leading to devices and applications that nobody has imagined yet."

The research of Parkins and his colleagues is outlined in their paper "Current Controlled Magnetic Domain-Wall Nanowire Shift Register”.  As is evident by the title, the current working concept of the memory utilizes the exotic carbon compound nanowires.  Fortunately, nanowires are a bit better researched and understood than quantum computing, and have seen limited commercial development.

On a technical standpoint the memory's operation is reliant on a phenomena in spintronics known as spin momentum transfer.  This allows for a magnetic shift register, which shifts a series of closely space domain walls.  Basically data is continuously pushed along through a nanowire.  The rotation of data through the looped circuit is analogous to the spin of the drive platter on traditional magnetic hard drives, albeit much faster, in that it brings the information to the read/write head.

Parkins states that magnetic domain wall data storage obstacles can be overcome by using spin polarized current along with the magnetized walls.  The resulting spin transfer torque will cause the domain wall to move.  He claims that the memory type is entirely new and never before considered.

The new memory type, like current flash drives would use no moving parts, limiting failure.  Further, as it stores information in electrons, IBM claims that the data would experience no degradation over time.  IBM brags that the drive could be rewritten “endlessly without any wear and tear.”

IBM believes it has the solution for the "How?" of improving memory densities and performance on hand.  Now the question they admit remains uncertain is "When?"  The researchers predict that in a decade the technology might see commercialization.  This seems overly optimistic over despite modest gains in quantum computing.

Still, one can only hope that IBM's new technology lives up to its bold statements and can be commercialized.  After all, it's good to look at things from a fresh perspective, and it would be even better to have one of those terabyte sized MP3 players, which the researchers are talking about, in your pocket.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

System Slowdown Imminent - Prepare Lifeboat!
By bldckstark on 4/11/2008 12:58:11 PM , Rating: 5
HDD's have become a joke in current computer systems. Compare them to just about everything else inside a computer today and it just doesn't make sense. CPU's have gotten thousands of times faster, as have GPU's, RAM sizes have grown a few thousand times and speeds have consistently increased. Optical storage mediums have gone from CD's to BR. HDD sizes have increased, but read/write rates have stagnated in comparison. The only part of a computer that has evolved less is the case.

Whoever actually creates a reasonably priced, robust, high speed replacement for hard drives as we know them today will rake in hundreds of billions of dollars. It would seem to me that we would have more alternatives to the current hard drive on the horizon considering the money to be made. Yet all we ever hear about is somebody who invented a system that we may never be able to manufacture and sell for a decent price.

But maybe it's just me.




RE: System Slowdown Imminent - Prepare Lifeboat!
By Schrag4 on 4/11/2008 2:06:33 PM , Rating: 2
Agreed. While hard drives have gotten much more reliable and somewhat faster over the last 10 years, they're still the thing in my system that I EXPECT to fail first. Not only that, but it's the only really 'important' part in anyone's system. Video card died? Buy a new one. Monitor died? Same answer. Hard drive failed? Oh boy, I hope you're still regularly making backups...


By puffpio on 4/11/2008 8:58:06 PM , Rating: 2
Well yeah..the HDD is one of the few remaining components that have moving parts..of course its gonna fail faster than things w/o moving parts


By Ananke on 4/11/2008 3:23:34 PM , Rating: 2
However, don't forget that is often feasible things to be patented, particularly ideas, since their range is broader. So, IBM will patent this research results, and license in several years to somebody like Intel, for actual implementation in memory chips.


RE: System Slowdown Imminent - Prepare Lifeboat!
By puffpio on 4/11/2008 8:59:49 PM , Rating: 2
I would dispute that because as HDD sizes increase, they are increasing densities, not adding more platters...therefore at the same RPM you get faster transfer speed since the data is stored more densely.


RE: System Slowdown Imminent - Prepare Lifeboat!
By Laitainion on 4/12/2008 4:47:53 AM , Rating: 2
But the scale of that increase is nowhere near the increases in CPU speeds, or even RAM speeds.

After all, DDR2 800 has a peak transfer rate of 6400MB/s whereas DDR 266 (which was standard roughly 10 years ago, I believe) is 2100MB/s. That's an increase of more than 3x.

On the other hand standard (i.e. 7200 RPM) hard drives are stuck on ~80MB/s I believe. Even if the increase in tranfer rate is similar to RAM over the last 10 years, the initial disparity means that hard discs are getting slower in proportion to the rest of the system.


By SlyNine on 4/13/2008 2:27:11 AM , Rating: 2
In 98 they were just moving from EDO ram.

http://www.youtube.com/watch?v=huyzCJvS6UQ Ignore the crappy Techno music at the start.

But it actually helps your point more then anything.


RE: System Slowdown Imminent - Prepare Lifeboat!
By CZroe on 4/12/2008 4:24:27 PM , Rating: 2
Hard drive densities have been scaling far faster than CPUs in the last 7 years. That also equates to sustained transfer and size scaling. CPUs aren't "thousands of times faster" when you compare apples-to-apples. In 1996, I bought my first X86 PC with a "P-rated" IBM P150 CPU (In actuality, a Cyrix 133MHz). It had a 1GB HDD and 16MB of RAM. My latest system has a Q6600 and 4GB of RAM with 3x1TB internal (2x500 OS, 2x1TB storage) and 1TB external. My CPU frequency is about 18x what my old one was, and you can multiply that by 4x if you must because that's still nowhere near "thousands of times faster" (about 72.1x). Even just one of my 1TB HDDs is exactly 1,000 times more than my original one. The farther back you go, the MORE extreme the ratio gets (disk speed and capacity scale faster). You wanna talk about something that isn't scaling fast enough? Removable optical media. 700MB CDs were once far bigger than any PC HDD, but even a prototype 200GB BD doesn't come close to a 1TB drive.

HDD random seek times stabilized when they hit their consumer limitations for 5400RPM, 7200RPM, and 10K RPM, so of course they won't scale. Similarly, HDD capacities will stop scaling when they hit their physical limitations sometime in the coming years, but it hasn't happened yet.


By Reclaimer77 on 4/12/2008 7:26:02 PM , Rating: 2
Wrong guys.

Maybe your just using the word " speed " as a very general term. But the speed of HDD's are not the problem. They are plenty fast. What makes them seem slow and lead to " system slowdown " as the OP put it, is the fact that it has to SEEK.

Thats why solid state drives are so exciting. Sure they are about as " fast " as current hard drives, but thats not the point. The seek times blow away hard drives, because there are no seek times for solid state.

This article is very good and exciting. And it should come as no surprise, because pretty much everything we take for granted in PC's today was in some part developed or based off technology developed by IBM.


By mathew7 on 4/14/2008 3:46:01 AM , Rating: 2
quote:
Maybe your just using the word " speed " as a very general term. But the speed of HDD's are not the problem. They are plenty fast. What makes them seem slow and lead to " system slowdown " as the OP put it, is the fact that it has to SEEK.


Defragmentation on HDD is needed BECAUSE OF SEEK TIMES (ok...it also helps in recoveries, but for that, backups are better/faster/more reliable). On a flash based device, it does not matter if you need the next sector or the next+1M sector, as the time is the same.
There is still the matter of multiple-block access, but the extra overhead of requesting 16 1-sector reads compared to 1 16-sector read is minimal on a SSD. Having 4Kb clusters makes (currently) 8-sector reads.


By SlyNine on 4/13/2008 2:36:47 AM , Rating: 2
Mhz per Mhz your C2Q would kill your Cyrix 133, I mean Murder.


By phxfreddy on 4/13/2008 8:30:06 AM , Rating: 2
Yah..its just you. Get an engineering and/or business degree and do it yourself if its so damned easy! ( its not or it would have been done! )


"We shipped it on Saturday. Then on Sunday, we rested." -- Steve Jobs on the iPad launch














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki