Non-volatile memory development gathers steam

November was a rather busy month in the development of non-volatile memory (NVM). A common thread within two announcements that will be considered here is the hybrid nature of memory solutions in the absence of the long elusive universal memory. The announcements might also suggest more modest goals for emerging memory, but that is not necessarily a bad thing.

There is plenty of opportunity to advance system performance in the absence of a universal memory. It is the latter of the announcements though that really brought home the notion that the comings and goings in NVM development. It has begun to look like an elaborate game of corporate musical chairs. Before the music starts, let's take a look at the technology.

In mid-November, Everspin announced the sampling of their first Spin-Torque (ST) MRAM.  

MRAM was of course one of the original three technologies touted as a candidate for the “Universal Memory”. While the 64 Mb density for this device is somewhat modest it should certainly be sufficient to service non-universal applications.  The Press Release in fact mentioned use of MRAM as a cache or buffer in solid state drives or RAID storage systems.  

Such an application was also being considered for FRAM, but it is unclear if FRAM can be produced at sufficient density. One common thread to come out of MRAM’s and FRAM’s recent histories is a focus on these non-universal applications that access specific properties of the memory to overcome or mitigate weaknesses in the performance of more traditional memories.  Solid-state drive (SSD) cache is probably a good example of this.
 
Moving on to the second announcement, Micron announced it will be collaborating with AgigA on the development of non-volatile DIMM technology.  

Very briefly, the release describes a “hybrid non-volatile RAM technology” that pairs DRAM and FLASH. In the case of power loss, an ultracapacitor provides power long enough for the data in the DRAM to be written to the FLASH. It is further noted that Micron will offer an NVDIMM to pair with an AgigA power module to “provide a complete NVDIMM solution”.

At the moment that is enough about the technology. What I found more interesting was the fact that AgigA is a subsidiary of Cypress. The mental wheels started turning and the music started playing.

Hold on a minute, I am not sure where to start.

Cypress is nearing the completion of its acquisition of Ramtron, which of course continues to develop FRAM. It was back in 2008 that Cypress acquired Simtek and their nvSRAM technology.  

What about Micron? They formed a relationship with Unity and their CMOx technology in 2011. Unity though was acquired by Rambus in February of this year.  

Micron on the other hand acquired Numonyx and their PCM technology in February 2010. Going further back Numonyx was spun out of Intel in April 2008. 


So, to say there is a bit of back and forth on the corporate side of NVM is an understatement. All of these movements are plotted in the Figure above.  

In this schematic, a company is depicted inside another if it is owned by or a subsidiary of the larger entity. If a company is shown adjacent to a larger entity there is some sort of working relationship, perhaps a collaboration, or maybe a development agreement.

We will just have to wait and see who is the last one sitting when the music stops. 


--Paul Boldt is a principal analyst at ned, maude, todd & rod Inc., an Ottawa-based technology research company.

[Editor's note: This blog first appeared in www.engineering.com. It was re-posted here with the author's permission.]