New memory technologies

Posted:
in Future Apple Hardware edited June 2015
There is a thread on seeking alpha regarding Intel and Micron's development of phase change memory. Stephen Breezy is predicting big things for Intel with what he considers the imminent release of PCM. He considers the technology so compelling a development that Apple would be forced to move their mobile devices back to X86 processors.

I have been following the new memory technologies from a distance as 3D NAND seems close to hitting a wall and the holy grail of volatile and non-volatile memory seems close to realization.

The technology that seems the most promising involves carbon nanotube memory for which Nantero seems to be close to realization. And the great thing is that they plan on licensing the technology. They have reduced the cost of production 10 fold over the past 2 years and the the chips can be built on current CMOS processes.

Stefan Lai who headed up the development of Intel's PCM program is now working as a consultant for Nantero. He made a statement regarding CNT (carbon nanotube) memory that is pretty compelling. It seems that at this point CNT memory is two years out from full commercialization. If Intel and Micron are ready to release a PCM based solution now, Mr. Breezy may have a compelling argument. But if PCM is not yet ready, I don't see how Intel's PCM based solution is superior.

Does anyone have any potential information regarding Intel and Micron's plans on PCM and whether they are getting ready to release an actual product to replace DRAM and 3D NAND?

Comments

  • Reply 1 of 4
    herbivoreherbivore Posts: 132member
    I realize that I made a mistake in the previous post. I meant to say the holy grail of volatile and non-volatile memory "convergence" seems close to realization.

    Instant booting, work never being lost in the event of a kernel panic, low power consumption, and the ability to scale are all very compelling.

    The following link describes the developments:

    http://www.computerworld.com/article/2929471/emerging-technology/fab-plants-are-now-making-superfast-carbon-nanotube-memory.html

    If Apple licenses the technology and incorporates the memory directly on the SOC, ARM based Mac OSX towers and laptops may not be too far behind. And it could lead to incredible advances for the watch.

    What do you all think?
  • Reply 2 of 4
    tallest skiltallest skil Posts: 43,399member

    I think we need to adopt SOMETHING new. Capacities have only increased 5x in almost ten years. The previous ten years, they increased 1000x.

  • Reply 3 of 4
    wizard69wizard69 Posts: 13,377member
    herbivore wrote: »
    There is a thread on seeking alpha regarding Intel and Micron's development of phase change memory. Stephen Breezy is predicting big things for Intel with what he considers the imminent release of PCM. He considers the technology so compelling a development that Apple would be forced to move their mobile devices back to X86 processors.
    If PCM actually works out for Intel they most likely would have to sell it to the market at large. I don't see them being able to keep it X86 only. Further I really don't see the tech as compelling especially in relation to other technologies in the pipeline.

    Consider this: it took flash decades to become the solid state technology of the era. I don't see these technologies taking over that quickly as it will take the market awhile to decide which is the best avenue to follow.
    I have been following the new memory technologies from a distance as 3D NAND seems close to hitting a wall and the holy grail of volatile and non-volatile memory seems close to realization.
    Convergence is an un-holy Grail if you ask me. It is an approach that is only workable in a limited number of applications. Beyond that I see no let up in the need for secondary store.
    The technology that seems the most promising involves carbon nanotube memory for which Nantero seems to be close to realization. And the great thing is that they plan on licensing the technology. They have reduced the cost of production 10 fold over the past 2 years and the the chips can be built on current CMOS processes.
    That will still take awhile to move into the mainstream market. I see two paths being pursued for at least the next decade. One will be better volition RAM technologies that we will see shortly with DDR4 and cube memories. The other is a bit more interesting actually and that is what will be the solution for secondary store moving forward. Here we will see a long shake out and frankly I'm not sure whom to back here.

    If I was to look at this seriously I'd have to ask which technology has the best growth potential when it comes to data density and operational speed. The reason I believe data density is so important is that I really don't see an end to the need for increasing secondary store capacities. I'd jump at the chance to get a 5TB SSD especially if it has the reliability figures that Nantero talks about. By the time that tech hits the market I will probably want a 10 TB disk. All at reasonable and competitive prices compare to today's technology.

    By the way I'm not a big believer in the cloud. In fact I kinda wish that Apple would get off its dead ass and give us real secondary store updates in its iOS devices.
    Stefan Lai who headed up the development of Intel's PCM program is now working as a consultant for Nantero. He made a statement regarding CNT (carbon nanotube) memory that is pretty compelling. It seems that at this point CNT memory is two years out from full commercialization. If Intel and Micron are ready to release a PCM based solution now, Mr. Breezy may have a compelling argument. But if PCM is not yet ready, I don't see how Intel's PCM based solution is superior.
    Ultimately the market will define what is superior for the masses. Interestingly Nantero will likely have nice markets for years to come due to their reliability in hostile environments. For us minions it will likely be a question of cost just as it is with SSDs today. The minute you can buy higher density and faster SSDs that aren't flash based you will see a new tech take over.
    Does anyone have any potential information regarding Intel and Micron's plans on PCM and whether they are getting ready to release an actual product to replace DRAM and 3D NAND?

    Nope! Haven't seen much lately in the trade magazines I have access to. The short term focus on volatile RAM is DDR 4 and the extremely fast cube technologies. For non volatile the focus seems to be on 3D flash technologies. Everything else at the moment seems to be wishful thinking or planning for ramp ups over the next 5 years. I've seen nothing that indicates a capacity replacement for flash yet and as noted capacity is very important.
  • Reply 4 of 4
    wizard69wizard69 Posts: 13,377member
    herbivore wrote: »
    I realize that I made a mistake in the previous post. I meant to say the holy grail of volatile and non-volatile memory "convergence" seems close to realization.
    We Understood.
    Instant booting, work never being lost in the event of a kernel panic,
    I'm not sure how this idea that these new technologies will result in not loosing work. By definition a kernel panic means that something is wrong with data or code and as such you have no assurance that you would get back to your data. In fact your data would be less secure in such a system, if it isn't backed up in secondary store.

    Plus such systems would require dramatically updated operating systems. Given that there are advantages especially for small scale embedded systems.
    low power consumption, and the ability to scale are all very compelling.
    These are huge factors and when the shake out comes I would expect that the winner would be the product that has the advantages here.
    The following link describes the developments:

    http://www.computerworld.com/article/2929471/emerging-technology/fab-plants-are-now-making-superfast-carbon-nanotube-memory.html

    If Apple licenses the technology and incorporates the memory directly on the SOC, ARM based Mac OSX towers and laptops may not be too far behind. And it could lead to incredible advances for the watch.
    Interestingly AMD and Intel are already showing the way forward here by putting very high speed RAM into the packages of the devices they sell. AMD is about to introduce GPUs with on package HBM, Intel already has built in high speed RAM on its high end APUs and they are also hard at work on the next Xeon Phi with memory in the package. The way forward is clear it is just a matter of how and when for Apple.

    By the way I know some of Apples "A" series chips have RAM in package, what I'm talking about here is a high performance option that is especially needed by the GPU. RAM performance is so important to the operation of today's APU type chips that I wouldn't be surprised at all to see the next set of "A" chips coming from Apple supporting DDR4, the long shot here would be Cube memory.
    What do you all think?

    I think the next few years will be very interesting. Short term 2016 I wouldn't be surprised at all to find all new devices coming from Apple supporting DDR 4 or variants there of. As far as secondary store goes flash will be with us for awhile, I just don't see anything competitive happening in 2016-17.
Sign In or Register to comment.