back to article Flash fryers have burger problems: You can't keep adding layers

The flash foundry folk took on 3D NAND because it provided an escape hatch from the NAND scaling trap of ever-decreasing cell sizes eventually to non-functioning flash. But 3D NAND, the layering of many 2D planar NAND chip structures, will run into its own problems. These are to do with wafer production time and yield, and …

  1. This post has been deleted by its author

    1. Steve K

      Re: "plasma breams"

      Yes, that sounds fishy to me too...

      1. Anonymous Custard
        Trollface

        Re: "plasma breams"

        I was also wondering how many layers would be required before they had to upgrade to sharks...

        1. PNGuinn
          Trollface

          Re: "plasma breams"

          "I was also wondering how many layers would be required before they had to upgrade to sharks..."

          Obligatory : With frikking lasers?

          1. Anonymous Custard
            Trollface

            Re: "plasma breams"

            There's another kind?

      2. Alistair
        Coat

        Re: "plasma breams"

        Would one pan fry that, or use it to heat the pan to fry the other piscean critters?

    2. TheVogon

      Re: "plasma breams"

      It's just scale up instead of scale out....

      1. TheElder

        Re: "plasma breams"

        Spending too much time on Twitter are we?

  2. Prst. V.Jeltz Silver badge
    Coat

    Wait , so this isnt about cooking burgers?

  3. Anonymous Coward
    Anonymous Coward

    we're running out of options

    Sure you can use TSV, but its expensive and that only delays the inevitable :

    Multi levels per cell - done ?

    lithography - done

    3D - done

    TSV die stack ....

    1. Alan Brown Silver badge

      Re: we're running out of options

      It's still denser and easier to do in the factory than have precision mechanical components at these kinds of scales out in the field.

      Going to smaller geometry on 3D is a non-starter because not only dies it reduce the number of write cycles and increase errors, but it makes the device significantly _slower_ - that was the other compelling reason to go back to larger geometry.

      Don't forget: we _already_ have 2.5" 16TB SSDs and the driving force on cost is availability, not the need to make more layers. Demand vastly outstrips supply and whilst you can add more production capacity, those etching machines don't come cheap.

      1. Anonymous Coward
        Anonymous Coward

        Re: we're running out of options

        Not everyone needs flash capable of 100,000 random 4K ops per second, or 3GB/sec of sequential read/write, or with a drive write per day for five years worth of longevity. For a lot of uses 1000 4K ops, 500GB/sec, and a drive write per week or even month is more than adequate, That provides a way to get cheaper flash for cold data, entry level PCs, cheap phones and so forth.

        If you can process 5x more terabytes of flash per fab line when making the denser stuff, it will cost a lot less since the equipment investment per TB is 1/5 of what it is on the faster longer life stuff.

    2. fpx

      Re: we're running out of options

      "Sure you can use TSV, but its expensive and that only delays the inevitable"

      Chip design and manufacturing has been pretty good at delaying the inevitable for like 30 years.

  4. Andy The Hat Silver badge
    Coat

    An idea ...

    Solder the devices on top of one another with only a free select pin to operate the low or high level (literally) device - double the capacity at probably 75% the performance ... with decent caching and a nifty algorithm possibly even better.

    If I suggested that was what we used to do in the 80's none of the youngsters would believe me ...

    1. Anonymous Coward
      Anonymous Coward

      Re: An idea ...

      Did that in my Amiga 1000. One advantage of my job was that a PACE micro-miniature repair station came with it. Perfect for these one-offs. Now, for serious points, did you ever use wire-wrap in your computer? Been there, done that.

      1. TheElder

        did you ever use wire-wrap

        Soldering iron on a Bendix G-15.

      2. td97402

        Re: An idea ...

        “Now, for serious points, did you ever use wire-wrap in your computer?“

        Same SYM-1 computer as in earlier post. Now with crude card cage and a home built RAM card, wire wrapped, 16KB more static RAM!

    2. td97402

      Re: An idea ...

      “If I suggested that was what we used to do in the 80's none of the youngsters would believe me”

      Exactly how I got my SYM-1 to 8KB when it only had sockets for 4KB of RAM.

  5. Anonymous Coward
    Anonymous Coward

    Why don't we

    Just teach people to properly manage their mailboxes properly and encourage the porn industry to use better video codecs.

    We'd save an incredible amount of space over night and negate the immediate need for more storage.

  6. Tromos
    Joke

    How many layers are too many?

    From a quick glance at the photograph, I would say take out all the veggie layers and substitute bacon for three of them.

  7. YARR
    Boffin

    Rather than electrically joining the layers, could each layer induce a current into the adjacent layers? This way the connection will work provided adjacent layers aren't offset by much, even if the entire stack doesn't line up perfectly.

  8. Conundrum1885

    possible

    Maybe. I did wonder if this might get around interconnect issues, provide chip power using 13.56 MHZ around the edge on each chip then offset chips similar to how microsd are made. Advantage is a bad shorted chip can simply be ignored not doom the whole tsv stack!

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon