Intel and Micron Announce "Revolutionary" Mystery Memory

Details on exactly what it is are vague

2 min read
Intel and Micron Announce "Revolutionary" Mystery Memory
What's are these memory bits made of? Intel and Micron aren't saying.
Illustration: Intel

This week, Intel and Micron announced 3D XPoint (“crosspoint”), a new form of nonvolatile memory that the companies say is 1000 times speedier than NAND Flash and ten times denser than DRAM. But what exactly is it? Good luck trying to figure it out.

“They’re being pointedly vague,” says Jim Handy, a memory analyst of Objective Analysis in Los Gatos, Calif. 

A press release on Intel’s website touts this as “the first new memory category since the introduction of NAND flash in 1989”. But I’m sure a number of companies would disagree with that characterization. There’s been plenty of work done on phase-change memory and other companies are pushing hard on resistive RAM. Everspin Technologies, a Freescale Semiconductor spin-off based in Arizona, has been shipping MRAM for years.

That said, Intel and Micron are big players in the semiconductor industry. Regardless of how unique 3D Xpoint is, their backing could really help launch alternative memory from the sidelines into mainstream adoption. 

Engineers have long hoped for a memory that could replace the mix we have now, something that could be speedy, dense, cheap, high endurance, and low power. Joel Hruska at ExtremeTech does a good job explaining how 3D Xpoint compares with existing memories based on the information currently available (he notes the 1000x-faster-Flash claim is somewhat vague but puts the new memory roughly on par with DRAM). 

There was a passing mention of “resistive elements” during a press conference on Tuesday, but changes in resistance are a common element in alternative memory technologies—that’s what denotes the difference between “0” and “1”. Still, some news outlets are betting, based on process of elimination and some other hints, that 3D Xpoint is some form of resistive RAM, which uses a voltage to alter the resistance of a material.

Intel and Micron say they plan to begin shipping the memory in 2016, so perhaps we’ll find out more about it then (their customers will likely find out more much sooner). Even if the companies themselves don’t release any more information, Handy says, “once they start releasing chips, everybody and his brother is going to know because reverse engineering is going to tear those things apart.”

Another big question is how 3D Xpoint will work its way into the existing ecosystem of memory and storage technologies. Handy says Intel could recommend it for its new server platform, details of which were leaked earlier this year and mention an alternate memory. Micron says it envisions the memory being used for both computation and storage.

The Conversation (0)

The Future of Deep Learning Is Photonic

Computing with light could slash the energy needs of neural networks

10 min read

This computer rendering depicts the pattern on a photonic chip that the author and his colleagues have devised for performing neural-network calculations using light.

Alexander Sludds
DarkBlue1

Think of the many tasks to which computers are being applied that in the not-so-distant past required human intuition. Computers routinely identify objects in images, transcribe speech, translate between languages, diagnose medical conditions, play complex games, and drive cars.

The technique that has empowered these stunning developments is called deep learning, a term that refers to mathematical models known as artificial neural networks. Deep learning is a subfield of machine learning, a branch of computer science based on fitting complex models to data.

Keep Reading ↓ Show less