Outside Thoughts:  The Importance of Specialized Memory Markets: Embedded NVM and OTP

Bob Merritt and Sherry L. Garber, Convergent Semiconductors

The forecast for the memory market always appears to center around a move to a universal memory that will replace all of the other memory types in the marketplace. This might be possible if end applications would quit expanding the range of memory performance attributes. Instead what is happening is a demand for a greater variety of memory types. One of those is Non-Volatile Memory (NVM), with demand that is growing exponentially as mobile devices with lower cost requirements proliferate. One of the stars is embedded NVM.

Current memory technologies support a value ratio between the processor and the memory that is heavily biased toward the processor.  This separation is essentially propagated by the fact that existing processes for high-performance memory and high-performance logic are incompatible.  With many new applications, higher performance processing and larger memory sizes are no longer the requirements.

Yet new and emerging memory technologies continue to focus on high-performance embedded NVM due to the potential process compatibilities.  The objective of these development programs is that for those applications for which the processing requirements are small enough, the memory performance attributes of density, speed, and non-volatility begin to contribute more and more of consumer's perceived value of the OEM's end product.  Ultimately this trend toward embedding high performance NVM in the same process with the processor will shift that value proposition toward the memory in many common applications with relatively simple processing requirements.  An example of this is that the relatively low performance of today’s Netbook processors is eroding demand from Laptops; processor suppliers are beginning to compete on cost, in other words manufacturing efficiencies, which had typically been the forte of memory technologies. 

Under the market conditions being supported by mobile personal computing devices, as the competitive value of the end product begins to focus more on the mobility and less on data processing, Moore's Law begins to assume a different impact on the value ratio between the various technologies.  Moore's Law suggests that in these highly mobile applications residing somewhere between cell phones and performance-oriented laptops, the single processor element becomes the commodity—the perceived value of the memory increases relative to the total user experience in terms of communications and stored entertainment-oriented capabilities.  Furthermore, once this ratio between the memory and the processing element begins to shift in the direction of the memory technologies, that point of transition in the value ratio between the memory performance attribute and the processor shifts upward in performance at a rate doubling every two years as predicted by Moore's Law. 

The decline of the desktop PC architecture as the primary target application for development of new memory technologies continues to increase the emphasis on other memory performance attributes. Trends in memory usage are providing a much richer selection of memory options and characteristics; a quick check with Micron’s website already shows over 1,128 different DRAM configurations separated into 23 distinct product groups.

However, it has been amply demonstrated that the exploration-driven selection of process variants and cell libraries can produce variances up to an order of magnitude for certain pivotal effects such as leakage. The ability to capitalize upon such staggering variances is uniquely available only at the architectural exploration stage; these variances cannot be achieved through the optimization process, since the decision that might have illuminated them has long since been locked-in. Furthermore, from an economic perspective, non-recurring engineering (NRE) costs and other costs associated with implementation can, depending upon chip volumes, outweigh the technical merits of such alternative solutions.

The demand for a wider selection of non-volatile memory technologies is likewise increasing at a rapid rate as we shift toward more mobile and customer-specific applications. The requirements for field programmability and customer-specific hardware customization of OEM products are continuing to accelerate.

The Sidense 1T-Fuse™ Logic NVM can be manufactured on standard logic CMOS process, requires no additional mask layers or process steps, and can be programmed in the field as well as during wafer or production testing. We believe that the market demand for such performance attributes is in the mainstream of memory applications, and that demand will continue to grow.

You can contact Bob and Sherry at Convergent Semiconductors:

Bob Merritt
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
949-488-7961

Sherry Garber
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
480-377-8020