IEEE Solid-State Circuits Magazine - Spring 2016 - 72

Price (US$/GB)

s
Co

ND
NA
ic
ntr
Ce

NOR

Speed

tric

A
DR

Density

Cost

n
Ce

CPU
SRAM

DRAM

SCM
NAND
PCIe SSD
SATA /SAS
SSD
HDD

NAND

Tape

Read?
Write?
NMT

~ms
~us
~ns
Performance Access Latency

~ps

FIGURE 6: Samsung's viewpoints of emerging memory.

of solid-state storage and the nonvolatile memory express (NVMe)
protocol, we know the tremendous
resources required to enlighten the
market on the benefits of a new technology. Further, as domain knowledge grows, so do the areas for
storage and device optimization and
education, making the supporting ecosystem a moving target.
When it comes to a new memory
technology, it is clear that a holistic
approach to ecosystem development
is necessary. IT industry stakeholders
should be supporting one another to
speed adoption and settle new technologies for the benefit of the entire
industry. There should be no question
as to the industry-agreed standard for
a new memory and interface, which
is why Samsung recently initiated a
discussion at JEDEC on protocols for
larger memory in emerging applica-

IOT

Gateway

tions such as big data analytics. (See
Figure 6.) We believe these applications could potentially be covered by
DRAM and NAND technology in the
near term and eventually by a new
memory technology such as highdensity PRAM.
Samsung is strongly committed to
leadership in both the scaling of current technologies and finding a new
memory technology. There are candidate technologies that may need
to be properly revisited as technology matures to be able to identify
as sustainable, viable commodity
memory technologies. Much like the
rise and widespread adoption of
solid-state storage took the previous seven to eight years, similar
changes with new memory technology will require seven to eight years
as well. Samsung hopes all industry
members can share in the benefits of

Network

FIGURE 7: Data flow in the IoT.

S P R I N G 2 0 16

IEEE SOLID-STATE CIRCUITS MAGAZINE

Data Center

the coming innovations and ecosystem development.

New Applications
Over the past decade, a huge amount
of data-"big data"-has been generated by widespread use of smartphones and lively communication
through social networks like Facebook and Twitter. However, as the
widespread deployment of the IoT
develops, data generation for the
next decade is expected to grow
exponentially. EMC estimates that the
IoT will generate tens of zettabytes of
new data, requiring significant innovations in both software and hardware to extract value from this huge
pool of information.
Figure 7 illustrates the flow of data
generated from "things" at the edge,
to local collection at the gateway,
transmissions through the network,
and arrival at the data center for more
valuable data analytics. First, "things"
are the edges of the system, where
physical information is sensed and
captured as data. Edge devices take
many forms but typically focus on low
cost rather than performance, excluding powerful processors and external memory or storage. Today, IoT
edge applications have widely differing requirements for performance,

Table of Contents for the Digital Edition of IEEE Solid-State Circuits Magazine - Spring 2016