Latest White Paper: New Memories for Efficient Computing

A Potpourri of Emerging MemoriesThere has been a lot of discussion in the trade press lately about new memory technologies.  This is with good reason: Existing memory technologies are approaching a limit after which bits can’t be shrunk any smaller, and that limit would put an end to Moore’s Law.

But there are even more compelling reasons for certain applications to convert from today’s leading technologies (like NAND flash, DRAM, NOR flash, SRAM, and EEPROM) to one of these new technologies, and that is the fact that the newer technologies all provide considerable energy savings in computing environments.

Objective Analysis has just published a white paper that can be downloaded for free which addresses a number of these technologies.  The white paper explains why energy is wasted with today’s technologies and how these new memory types can dramatically reduce energy consumption.

It also provides a mini review of a number of currently-available or upcoming new memory alternatives including:

  • Resistive RAM (ReRAM) chips like Crossbar, oxygen vacancy, and conductive bridge technologies
  • Magnetic RAM (MRAM), both flash-like and SRAM-like
  • Ferroelectric Memory (FRAM) including PZT and halfnium oxide
  • Phase-Change Memory (PCM) a category into which we sweep the 3D XPoint Memory

Something consistent about all of them is that they are nonvolatile, so they don’t need to be refreshed like DRAM, they use faster and lower-energy write mechanisms than either NAND or NOR flash, and their memory cells can be shrunk smaller than current memory technologies’ scaling limits, which means that they should eventually be priced lower than today’s memory chips.

Have a look – it’s free.  The Memory Guy always finds such a price to be compelling!

4 Responses to Latest White Paper: New Memories for Efficient Computing

  • Lu Tseng Fu says:

    Thanks for the sharing. It’s useful to see the difference between the candities of emerging memory.

    • Jim Handy says:

      Thanks, Lu Tseng Fu!

      This has been a very interesting field for a very long time. I hope that soon these new memories will see more success than they have experienced in the past.

      Jim

  • I am sorry to see that all of these so-called Low Energy switches are not examples of adiabatic or Quasi-adiabatic switching (QAS). For low dissipation, the switching should be a lot slower than the actual transition time intrinsic to the switching phenomenology. A closed one would be STTRAM. That is, the intrinsic switching time (flipping spin or transfer) could be seen as QAS because of the W/R pulse is slower than this intrinsic “RC-like” time. The “Adiabatic Theorem” should be the theoretical leading principle to focus on achieving low energy. As we all know, adiabatic means that the energy-coupled into switching does not cause dissipation. One clear example is a Quantum Phase Transition, like the “True” Mott transition. The change from Insulator to a metal and vice versa, happens in time scales of tens of Femtoseconds, way below the typical clock pulse width. In the case of the “Mott Transistor”, this is messed up by the read/write mechanism being dependent on charge transport in the channel, so that dissipation due to scattering still follow semi-classical Ohm’s (Drude) law. However, the switching from one conductivity level to another, inside the Mott insulator due to “Field-induced Dynamic electron doping (Band Filling)” is QAS. So, we waste the low energy by the sensing method. From today on, see the DARPA/Applied Materials Press release, to see how CeRAM (not in your picture) does this the right way.

    • Jim Handy says:

      Carlos,

      Good to see your comment!

      Yes, the Symetrix CeRAM is a ReRAM type that I see was recently announced to have won a DARPA contract through Applied Materials and ARM.

      Congratulations!

      Jim

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.