Slashdot Mirror

← Back to Stories (view on slashdot.org)

IBM's Optical Storage Is 50 Times Faster Than Flash, And Also Cheaper (prnewswire.com)

Posted by msmash on from the beaming-its-way-in dept.

Flash storage is not as fast as the main memory (RAM); but RAM can't be used to store your regular files because of its volatile nature (and also because it's expensive). It appears we may soon have the perfect middle ground of the two. Scientists at IBM have demonstrated reliably storing 3 bits of data per cell using a relatively new memory technology known as phase-change memory (PCM). Engadget reports: To store PCM data on a Blu-ray disk, you apply a high current to amorphous (non-crystalline) glass materials, transforming them into a more conductive crystal form. To read it back, you apply a lower voltage to measure conductivity -- when it's high, the state is "1," and when it's low, it's "0." By heating up the materials, more states can be stored, but the problem is that the crystals can "drift" depending on the ambient temperature. IBM's team figured out how to track and encode those variations, allowing them to reliably read 3-bits of data per cell long after it was written. That suddenly makes PCM a lot more interesting -- its speed is currently much better than flash, but the costs are as high as RAM thanks to the low density.

47 of 77 comments (clear)

  1. So it's cheaper by goombah99 · · Score: 5, Insightful

    in the sense that it costs more?

    --
    Some drink at the fountain of knowledge. Others just gargle.
    1. Re:So it's cheaper by fishscene · · Score: 2

      That *is* odd... but perhaps it's because they just can't even.

    2. Re:So it's cheaper by rahvin112 · · Score: 3, Interesting

      About the same as claiming it's an optical format when it's actually phase change. Throw those buzz words in.

      Phase change memory already exists as Intel/Micron XPoint memory which should go on sale later this year (it already been sampling for a few months).

    3. Re:So it's cheaper by DoofusOfDeath · · Score: 2

      in the sense that it costs more?

      They make up for it in volume.

      Math!

    4. Re:So it's cheaper by rahvin112 · · Score: 1

      Demonstrating something in a lab and in a factory are completely different. You might be able to build MRAM in a lab at a million dollars a megabyte but until it can be mass produced at consumer prices it shouldn't be compared to things that are.

      What's held up the vast majority of these technologies is the cost and ability to mass produce. They invent new stuff all the time then they find out they can't mass produce it or that it will cost so much it won't sell. It's great that IBM figured out a roadblock, but you won't see any mention of cost or mass reproduceability because they haven't even looked at that.

  2. Whoa. by charlesbakerharris · · Score: 5, Funny

    That must make it, like, 500 times faster than Silverlight.

  3. Tripple Redundancy by Anonymous Coward · · Score: 1

    "amorphous (non-crystalline) glass"

    So a glass, glass-glass then?

  4. Re:Get your facts straight... is it cheaper or not by crunchygranola · · Score: 3, Insightful

    In addition, phase change memory is not "optical". Cr@ppy summary, even for /.

    --
    Second class citizen of the New Gilded Age
  5. Ternary Digits by Art+Challenor · · Score: 2

    Finally! Ternary Digits, now I won't be twiddling bits all day.

    1. Re: Ternary Digits by Crashmarik · · Score: 1

      I've always dreamed of getting paid for that.

    2. Re:Ternary Digits by phoenix_rizzen · · Score: 1

      So you'll be biddling twits all day instead?

    3. Re:Ternary Digits by fnj · · Score: 4, Funny

      So you'll be biddling twits all day instead?

      You messed it up. It goes like this: if binary digits are bits, ternary digits must be tits. Twiddling tits. That's what he'll be doing.

    4. Re:Ternary Digits by omnichad · · Score: 2

      It's Octonary. There are 8 states per cell, meaning 3 bits of storage. Unless it's just a really badly written article.

    5. Re:Ternary Digits by Art+Challenor · · Score: 1

      I hate when the facts ruin a good joke!

    6. Re:Ternary Digits by Christian+Smith · · Score: 1

      So you'll be biddling twits all day instead?

      You messed it up. It goes like this: if binary digits are bits, ternary digits must be tits. Twiddling tits. That's what he'll be doing.

      Is this a bad time to mention hard disks?

    7. Re:Ternary Digits by phoenix_rizzen · · Score: 1

      No, you missed the word play involved.

  6. Article is completely misleading... 1960's tech by PublicSchill · · Score: 1

    This has been researched since the 1960s, but it just was never viable for commercial production. Yes, this is a mojor breakthrough, but it's facing heavy competition. Intel/Micron's new memory tech will be out and it has already been demonstrated to be 1000 times faster than flash.

  7. Not optical, not fast, not 3DXP by duckintheface · · Score: 5, Informative

    Despite the use of the term "optical" there is nothing optical or "blue-ray disk" about this. Calcogenide glass is the PCM material. It is written and read with the application of voltage. There is no spinning disk involved.

    The most obvious omission is a comparison to the Micron/Intel 3D Cross Point memory announced last summer and scheduled for commercial introduction in 2017. 3DXP is 1000 times faster than flash (not just 50 times faster). There would also seem to be a number of patent issues since 3DXP also uses calcogenide crystals as the storage medium.

    It is journalistic malpractice to write an article like the two linked here without comparing the IBM research to the previously announced work by Micron and Intel.

    --
    "He took a duck in the face at 250 knots." -- William Gibson, Pattern Recognition
    1. Re:Not optical, not fast, not 3DXP by wile_e_wonka · · Score: 2

      Can I sue for "journalistic malpractice"?

    2. Re:Not optical, not fast, not 3DXP by NatasRevol · · Score: 1

      Well, Trump seems to want to, so sure why not?

      --
      There are two types of people in the world: Those who crave closure
    3. Re:Not optical, not fast, not 3DXP by Anonymous Coward · · Score: 1

      Can I sue for "journalistic malpractice"?

      You'd first have to prove that there are actual standards for "journalism".

      Good luck with that...

    4. Re: Not optical, not fast, not 3DXP by jd2112 · · Score: 3, Funny

      I once suggested that for Earth Day, the St. Louis Post Dispatch could state they are going green by wrapping copper wire around the corpse of Joseph Pulitzer (founder of STLPD in addition to having journalism award named for him) and powering their printing presses by him spinning in his grave.

      --
      Any insufficiently advanced magic is indistinguishable from technology.
    5. Re: Not optical, not fast, not 3DXP by jd2112 · · Score: 2

      I once suggested that for Earth Day, the St. Louis Post Dispatch could state they are going green by wrapping copper wire around the corpse of Joseph Pulitzer and powering their printing presses by him spinning in his grave.

      --
      Any insufficiently advanced magic is indistinguishable from technology.
    6. Re: Not optical, not fast, not 3DXP by Anonymous Coward · · Score: 1

      Not just once, in fact.

    7. Re:Not optical, not fast, not 3DXP by pipingguy · · Score: 1

      "It is journalistic malpractice..."

      Well, first we would have to have unbiased journalists who have wide and deep knowledge of their subject matters, are not beholden to "editorial content standards" and not ideologically or financially bound to...I think I see the problem.

  8. 3d X-point memory by Anonymous Coward · · Score: 1

    The 3d xpoint resistive memory being developed has higher density than flash, should be manufacturable with modest changes to exising fab methods, has no write life limitations, is way faster than flash, does not need to be updated in large blocks like flash, does not use exotic materials, and should be cheaper than flash once mature.

    Why are we considering physical phase change materials that have potential write life issues from crystal grain migration/unexpected crystal growth over the cell domans, when we have functioning prototypes of a vastly superior technology, just looking for backers?

    This may be useful for new optical discs, but xpoint will make that obsolete too.

    1. Re:3d X-point memory by mark-t · · Score: 2

      Just because something doesn't run out, doesn't mean that you never need more of it.

      Ask any serious Lego enthusiast.

      --
      File under 'M' for 'Manic ranting'
  9. Real announcement by PublicSchill · · Score: 1

    Website about the research(note, there is a list of papers too... this "news" is actually from 2015) https://www.research.ibm.com/l... Youtube video: https://youtu.be/q3dIw3uAyE8 And yes... they only have a prototype. There are still lots of possibly, maybe, somedays in this announcment.

  10. Confusing Summary by DavidMZ · · Score: 2

    PCM is not limited to re-writable Blu-Ray, it is actually used in memory chips.

    Micron used to manufacture PCM memory chips and dropped them in 2014. There are also some debate regarding whether IMFT's 3D XPoint is also PCM or not.

    The real innovation in IBM's work is turning PCM into a TLC, and that is really impressive.

    1. Re:Confusing Summary by friesofdoom · · Score: 1

      PCM in blu ray discs stands for "Pulse Coded Modulation" while PCM in memory chips stands for "Phase-Change Memory". These are not the same PCM, whoever wrote the summary is very very confused.

    2. Re:Confusing Summary by DavidMZ · · Score: 1
      Thanks for the precision.

      To add to the confusion, re-writable blu-ray disks actually use Phase Change materials.

  11. What tier does this storage belong? by mlts · · Score: 1

    I am not sure what tier this storage goes in, as for price/capacity, where CPU registers are the most precious and tape, cloud, or big, slow DASDs are on the other scale of speed/price.

    Does it go under flash, but "above" spinning rust? The BD example made it seem that IBM is wanting to make another CD-PD, which was a storage format that lasted a few years until CD-R and CD-RW became mainstream, where a drive could read CDs, as well as use a specific optical cartridge.

    Personally, I would love a high density, cheap optical storage medium. Optical disks are easier to manipulate than tapes (wasn't that long ago when everyone had 400 CD autochangers), and because of the size, can be stored with more units in a given space than tapes. If the optical medium is cheap and WORM, that would provide excellent protection against ransomware, as well as the ability to store media offline.

    If it goes between flash and spinning disks, that isn't really a useful market niche, as flash is getting cheaper all the time.

    1. Re:What tier does this storage belong? by swb · · Score: 1

      I get the impression that this and IntelMicron's 3DXpoint are really meant to be replacements for current uses of SSD and most uses of disk.

      I have the vague feeling that maximal price extraction will see 3DXpoint used in some cases as a caching layer between CPU and flash disk initially until manufacturing capacity ramps up and the production cost gets low enough to drive SSD flash out of the market, which it ought to based on performance alone.

      Hopefully it will drop in price enough to kill off hard disk completely, SSD seems like its on the way there now for most use cases.

      It would be nice to see an archival media oriented at home users -- something like LTO-6 levels of density and form factor, but with a friendlier interface like USB3(.1). Honestly, I have a hard time seeing why you couldn't run an LTO-6 drive off USB3 now. But the tapes and drives are awful expensive for non-enterprise use.

    2. Re:What tier does this storage belong? by DavidMZ · · Score: 1

      PCM won't kill flash (3D NAND has some cost benefits), it will displace DRAM in some applications.

      And I would not count HD out as well, there has been some impressive advances published lately by SanDisk and the like.

      Computerworld published recently a nice article on the Memory landscape

    3. Re:What tier does this storage belong? by mlts · · Score: 1

      I have been asking the same thing. The LTO drives have already dealt with shoe-shining, so they can work on a sub-optimal interface, and even then, there is always having the drive with an I/O buffer similar to CDs before "burn-proofing" made buffer underruns a thing of the past.

      I do know LTO-3 drives can run from Thunderbolt 1, which is about on par with USB 3, so it can be done. It is just a matter of making consumer level software to handle LTFS, and some basic documentation, like not to run a defragmenter on a tape volume.

      Get tape drives to a $1000 price point, easily used with USB 3.1, and this definitely would be a storage niche, especially with ransomware. A WORM tape or optical media is as good as it gets, no matter what gets compromised. If tape companies do this, they definitely will have an expanded market, especially if their product is combined with a NAS of some sort, so a backup of the NAS can be as simple as putting tape in and hitting a button.

    4. Re:What tier does this storage belong? by swb · · Score: 1

      The thing is USB3 and 3.1 really aren't sub-optimal interfaces. USB 2 was, but 3 eliminated the interrupt problem and took it to 5 Gbps and 3.1 to 10 Gbps.

      With this kind of bandwidth, I would kind of hope manufacturers would look to expand market penetration on some "enterprise" devices and realize that between the inexpensive, high speed connectors and operating system device virtualization that more devices could be effectively connected via USB3 without SCSI as a dependency.

    5. Re:What tier does this storage belong? by swb · · Score: 1

      I think PCM will have to cut costs to become price competitive with 3D NAND as primary storage in order to survive. I don't think it's cost or value proposition as some kind of buffer technology or DRAM augmentation will get it very far.

      For better or for worse, the entire technology field is stuck in a DRAM/disk paradigm and no emerging storage technology (as in likely to be mass produced for consumption in 5 years) will be able to unify CPU and disk storage. If 3DXpoint lives to 100% of its hype, it might be useful for slow low power devices, but mostly you're talking about marginal gains at the extremes of low power consumption to cut out DRAM completely or reduce it.

      And even if one unifying storage tech turned up that could, it would take years for operating systems, applications and users to restructure how they thought about and used a unified storage paradigm.

  12. Needs more buzzwords! by DidgetMaster · · Score: 1

    It sounds cool, but how does it incorporate lasers and virtual reality while traveling at warp speed?

  13. I told you guys PCM was going to be here. by Khyber · · Score: 2

    One of the few techs to actually hit market within about 10 years of being proven feasible.

    It's good stuff, but this stuff is slow in comparison to Intel/Micron's stuff.

    If IBM partnered with AMD on making a 3D stacked version of this, they could easily catch up or even be faster, with higher bandwidth to boot.

    --
    Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
  14. Re:Middle ground? by fluffernutter · · Score: 1

    Because this memory is 3x as dense as RAM by storing three bits in a cell, thus making it allegedly 1/3 the price of RAM.

    --
    Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
  15. Re:Failing to see what part is "optical" by MightyYar · · Score: 3, Funny

    You can see it.

    --
    W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
  16. Summary, and possibly the story, are nonsense. by friesofdoom · · Score: 5, Informative

    "To store PCM data on a Blu-ray disk, you apply a high current to amorphous (non-crystalline) glass materials, transforming them into a more conductive crystal form. To read it back, you apply a lower voltage to measure conductivity -- when it's high, the state is "1," and when it's low, it's "0.""

    1. We do not write to blue ray disc by applying a voltage, we shine lasers at it.
    2. The summary seems to have confused the fact that Chalcogenide is used in both RW Optical media, and phase-change memory.
    3. The summary thinks that IBM has invented a new optical memory, when they are clearly talking about phase-change memory.
    4. The summary has confused two acronyms of PCM - Phase Change Memory and Pulse Coded Modulation.

    This whole summary is utterly cringe-inducing and complete garbage. Sorry.

  17. Like to have non battry backed ram temp disk by Joe_Dragon · · Score: 1

    Like to have non battry backed ram temp disk that does not need to even have all matched speeds.

    That can make good use of all older ram that people have laying around.

  18. Re:Middle ground? by omnichad · · Score: 1

    costs are [currently] as high as RAM thanks to the low density

    Once this reaches market, that will change. The "current" cost of this prototype isn't really part of the equation.

  19. Re:Get your facts straight... is it cheaper or not by omnichad · · Score: 1

    The laser control on a BD-R Blu-Ray writer uses PWM. Change the W to a C and it's PCM. Then somehow it becomes this new thing.

  20. Re:Get your facts straight... is it cheaper or not by PublicSchill · · Score: 1

    So uh, in summary... Slashdot is good for nothing. Even if you did not terrible advertisements and worthless commentary, you're better off with reddit, or god forbid... some conde-naste "tech news" site. I heard they completely destroyed Arstechnica(RIP)

  21. Memresistors are going to make this all obsolete by TheNarrator · · Score: 2

    HP's memresistor project "The machine" is going to so thrash this technology to hell and back. They are talking about petabytes of data as fast as ram. You won't need to have a separation between ram and disk anymore. They already have prototype and they should show up in a few years.

    http://fortune.com/2016/03/02/...