|new market opportunities
and technical possibilities for flash as volatile memory |
December 3, 2015- The split personality of the future flash market - due to
emerging uses of flash as replacements for server
DRAM (a role which
de-emphasizes the non volatile characteristic of flash) is one of the ideas
discussed in my new home page blog on StorageSearch.com -
the big SSD
ideas of 2015.
TrendFocus compares worldwide raw physical storage capacity of
flash and HDDs
Editor:- October 5, 2015 - A new blog by TrendFocus -
far does NAND output have to grow in order to supply all our storage devices?
- says that that 80EB of NAND flash will ship this year compared to 500EB
The author Don Jeanette
concludes - "it is evident that there is not enough NAND supply to take
over all the storage requirements in the world at this point."
comments:- that's true as far as it goes.
But in my classic
article - meet Ken
- and the enterprise SSD software event horizon (2013) - I explained why I
think that SSDs will easily replace all hard drives in the enterprise
much sooner than this type of capacity gap comparison would lead you to
think. (It's a system architecture and virtualization thing.)
talliers and storage market research directory
RRAM SSDs in 2016? - Crossbar gets $35 million series D funding
September 14, 2015 - Crossbar
it has completed a $35 million Series D funding round bringing total
investment to $85 million to date.
Crossbar plans to use the funds to
continue the commercial ramp of its
RRAM NVM memory
technology which is based on a simple device structure using CMOS friendly
materials and standard manufacturing processes. It can be stacked in 3D, making
it possible to combine logic and memory onto a single chip at the latest
Crossbar is currently working with beta customers to
bring products to market in 2016.
Intel, Micron 3D ReRAM
Editor:- August 24, 2015 -
July Intel and Micron
a new bulk material based resistive memory nvRAM platform which they called
XPoint technology (later branded as Optane). At that time - the
technical information about the memory technology were vague and lacking in
More details emerged during the shows which immediately
followed (FMS and IDF) and here's a link with the
says cost per bit is likely to be somewhere between DRAM and nand flash.
is said to be 1,000x faster than nand but slower than DRAM.
density? A single chip can store 128Gb.
Sampling? Later this year with
production in 2016.
Some of the many form factors and attach points
which might benefit from this new technology are PCIe SSDs and Memory Channel
As with any new memory technology it will take time and
experience to prove whether Optane memory has enterprise grade reliability. For
this reason and due to the need to establish a new software ecosystem - early
uses of the memory will probably be in experimental cloud appliances and
consumer gaming devices.
...Later:- Initially I had serious
doubts about the market readiness state of the Intel / Micron preannouncement
because it appeared to leapfrog previously known memory offerings. And
storage history has
taught us 2 valuable lessons about new memories.
- the new memory is usually a small increment (2x, 4x etc) what was done
before - to minimize the risk of new problems creeping into the next scaled
geometry iteration, and
Where can you find more
reliable information about ReRAM?
- I've heard such "market breakthough stories"
from the anti-flash nvm world many times in the past 12 years - usually
precipated by a need for more investment cash.
I've found a website which seems to
have a more measured and informed approach to what has been happening in ReRAM
land - and reading it may help you guess better when these advances might
really intersect with the mainstream SSD market.
Take a look at
Avalanche Technology samples 64Mb STT-MRAM made using 55nm CMOS
Editor:- July 1, 2015 -Avalanche Technology
it is sampling the industry's first STT-MRAM chips manufactured using
standard CMOS 300mm wafer processing.
Avalanche's new memory device
is a 64Mb chip with an industry standard SPI interface built on a 55nm node
Nantero gets $31 million funding for 300º C retention
Editor:- June 2, 2015 -
a $31 million Series E financing round for its
NRAM technology which the company
says is scalable to below 5nm and which has >1,000 years retention at 85º
C or more than 10 years at 300º C.
Nantero was founded 14 years ago, and the last time I wrote about them was in
But the size and educational sophistication of the SSD ecosystem
today means that designers
(and investors) can
appreciate the nuances of difference which might be useful in extreme boundary
Offering a scalability roadmap below the current
commercial limits of flash,
and ruggedness way beyond flash - Nantero's technology has attractive features
which might lure SSD designers out of their 40 year comfort zone of
trapped charges in semiconductor cells.
Micron in production with 2D 3 bits per cell 16nm nand flash
June 2, 2015 - 2 years after sampling its
16nm nand flash - which was 2D with MLC nodes (2 bits per cell) -
it has progressed to the next evolutionary step and is now shipping 16nm
(which is still 2D) but is now 3 bits per cell (TLC).
In both cases
the products were 16GB memory chips.
Micron says it believes that
TLC will account for almost 50% of the total NAND gigabytes shipped in 2015.
update on the readiness of non flash NVMs to participate in SSDs
May 28, 2015 - In various interview clips in a recent article -
NAND, MRAM, RRAM: Emerging opportunities and challenges in
Solid State Technology - the author Paula Doe
reports how some of the contenders to
flash memory see their
roles within the SSD ecosystem. For example:-
- "Demand for ST-RAM is coming from buffer storage applications, such
as high-end enterprise-class SSDs..."
Some of these applications have been intuitively
obvious for some while - but this article gives a better idea of commercial
readiness and an indication of whether the next generation problems are being
tackled in a fast enough timeframe to be relevant to the
SSD market. ...read
- ReRAM has already been promised for delivery in
military SSDs (Jan 2015 news) but
forthcoming advances in repairable vertical architecture could increase the
desnsity to the point where it's attractive as an intermediate level of memory
in servers too...
How much 3D flash in 2015?
Editor:- May 5, 2015 -
that 3D will make up just 7% of NAND flash's average annual output for
who's who in ReRAM? - IHS article
Editor:- May 1,
2015 -Who's doing what re the commercialization of
- one of the seldomly heard from
NVM cousins - can be
learned in a new article -
Embedded ReRAM to 28nm - written by Peter Clarke
which appeared in IHSElectronics360.
other things re ReRAM - Peter Clarke says - "It has been the subject of
much research over the last decade because it had been predicted that NAND
flash memory would fail to scale beyond critical dimensions of 20nm."
article tells you which companies are still in this technology and discusses
current memory densities and controllers. ...read
samples 48-layer 3D nand
Editor:- March 26, 2015 - Toshiba today
it is sampling the world's first 48-layer 3D stacked 2 bit nand
flash memory in 16GB
chips aimed at the high capacity SSD market.
Mass production is
anticipated to be in the first half of 2016.
Intel and Micron promise 32 layer 3D nand SSDs by 2016
March 26, 2015 - Micron
it is sampling a new 32 layer 3D nand flash memory using floating gate
cells - which has been designed in collaboration with Intel - and which
provides 32GB MLC (2 bits per cell) in a single chip.
density TLC (3 bits per cell) version with 48GB capacity will sample in the
next few months.
Both devices are expected to be available in SSDs
within the next year.
3D InCites blog re Samsung's 3D TLC
Editor:- March 4,
2015 - What happens when you combine
bits per cell in the same flash?
A recent blog -
V-NAND Flash at the 2015 ISSCC published on 3D InCites summarizes the key
parameters of Samsung's
approach to combining 3D and TLC and offers some critical analysis.
on the directions for future advances - the blog's author Andrew Walker
says "I also heard that they may be looking at 4 bits/cell."
V-NAND page, Unveiling
XLC Flash SSD Technology (March 2008 )
Western Digital invests in Skyera's MRAM supplier
January 26, 2015 - Western
Digital's investment unit was among the investors in a $29 million
series B funding round in Everspin Technologies
President and CEO of Everspin said "With a leading worldwide foundry and
storage customer participating in Everspin's Series B investment round, the
entire industry spectrum is acknowledging ST-MRAM as the leading contender to
drive beyond the limits of current mainstream memory."
comments:- Everspin's MRAM is
tier of the non volatile caching technology used in
Skyera - and my guess is that this investment in Everspin is to take out some of
the risk of future availability of these memory parts at a time when an assured
supply at higher volume may soon be needed.
So you want x3 and 3D?
Editor:- January 23, 2015 -
Even if you already thought that
and DSP was an essential way for getting usable SSDs out of smaller 2D nand
flash - then there are even more reasons for using this technology on the
journey into 3D.
That's the conclusion you'll come away with after
seeing a paper (presented by DensBits at the 2014
Flash Memory Summit)
Necessity for a Memory Modem in 3D Memories (pdf)
things in this paper:- DensBits says that the scope for inter-cell interference
grows from 8 identifiable routes in 2D to 26 for each cell in 3D.
memory modem technology (DensBits's branding for their collection of adaptive
R/W DSP IPs) will (over and above everything it already does for 2D)
intelligently decouple read operations according to the severity of read
operations expected in the new 3D architectures - and even supports the notion
of TLC (x3) within 3D. (Which "needs state of art decoder and signal
Their conclusion? - Memory Modem technology is
required for 3D NAND scaling ...read
data integrity in
renewing the spin on MRAM's bright future
December 18, 2014 - Everspin
Technologies said at a recent event they have shipped over 40
million MRAM devices.
That's one of several interesting observations on the state of the
MRAM and RRAM market contained in a new blog -
Computers that Don't Forget - by Tom Coughlin,
President - Coughlin
Associates - who predicts that the market for MRAM devices may exceed
$2 billion by 2019. ...read
Cypress merges with Spansion
Editor:- December 1,
2014 - Cypress
Semiconductor and Spansion today
announced a definitive
agreement to merge in an all-stock, tax-free transaction valued at approximately
"...Our combined company will be a leading provider
of embedded MCUs and specialized memories" said T. J. Rodgers, Cypress's founding
president and CEO.
SSDs are made of this
Editor:- October 14, 2014 -
Without memory - there would be no SSDs.
And while naturally the
emphasis in SSD thinking is mostly on - how can we do useful and affordable
things with SSDs? - despite how terribly flawed the
raw material is which we
have to work with (which leads you to
and software) - it can
nevertheless be strategically useful for SSD specifiers to sometimes brace
themselves for a deep dive down into the cold details of how much better (or
worse) those raw memory characteristics are going to get - so you can
anticipate future developments.
This week the best place to look is
We need new software abstractions to efficiently handle all the
different emerging flavors of persistent enterprise memory - says SanDisk
October 3, 2014 - New enterprise software abstractions are needed in order
to efficiently utilize
all those unruly developments in
And laying the
framework for those ideas - along with some practical suggestions for where
applicable solutions might be coming from - is the theme of a recent blog -
Emergence of Software-Defined Memory - written by Nisha Talagala,
Fellow at SanDisk
- who (among other things) says:-
"We're seeing excitement build
for a new class of memory:- persistent memory - which has the persistence
capabilities of storage and access performance similar to memory.
"Given this richness of media technologies, we now have the
ability to create systems and data center solutions which combine a variety of
memory types to accelerate applications, reduce power, improve server
consolidation, and more.
"We believe these trends will drive a
new set of software abstractions for these systems which will emerge as
software-defined memory a software driven approach to optimizing memory
of all types in the data center." ...read
See also:- are you ready to
rethink enterprise DRAM architecture?
Micron's enterprise SSD revenue grew 79% QOQ
September 25, 2014 - In its Q4 earning conference call today Micron said that about
66% to 75% of its nand flash had gone into client SSDs - with the remainder
being enterprise. However Micron also said its enterprise SSD revenue was up
79% quarter-on-quarter. ...full
transcript on SeekingAlpha.com
STT-MRAM? - update report
Editor:- July 18, 2014 -
IEEE Spectrum today
published an interesting state of the art article about spin-transfer-torque
Memory Shows its Might.
Among other things the article's author -
Editor-in-Chief of IEEE Spectrum Magazine - says "STT-MRAM may be claiming
some of the enthusiasm once reserved for other alternative memories, such as
ferroelectric RAM, phase-change memory, and resistive RAM. But its success will
come down to manufacturing technology and how well it can compete on cost."
SanDisk and Toshiba collaborate on 3D nand fab
May 13, 2014 - SanDisk
and Toshiba today
that they have begun work on demolishing and converting a 2D NAND fab at
Yokkaichi Operations, in Mie prefecture, Japan over to 3D capability with a
view to enabling 3D output in 2016.
Samsung starts 3D nand production at new fab in China
May 9, 2014 - Samsung
that its new memory fabrication line in Xi'an China - which will make 3D
V-NAND - has begun full-scale manufacturing operations.
50% of global
NAND flash is made or processed in China.
I just wanted solid-state memory at a cost per bit as low as a
CD-ROM or a DVD - said Contour Semiconductor's founder - whose company
yesterday named a new CEO
Editor:- April 23, 2014 - Contour Semiconductor
is a new (long time in development) company which I only learned about this
week via a couple of my linkedin contacts.
You might want to learn
more about them too.
"Contour's new chip
technology has the potential to be every bit as disruptive to the solid state
flash market as flash was to hard disks drives" says Saul Zales who
Contour's new CEO in a press release yesterday.
Saul Zales is well
qualified to judge those markets - as his background includes flash or SSD
related business development at some well known SSD companies - namely
3D NAND flash challenges - an industry roundtable discussion
Editor:- February 6, 2014 - The best article I've yet seen about
the practical implications of increasing the adoption of 3D NAND flash is -
At The Table: Commercial potential and production challenges for 3D NAND memory
technology published by Semiconductor
Manufacturing and Design.
Among the many practical
considerations discussed in this article was the question of - "how is the
semi industry preparing for the transition to 3D memory?"
issue of scalability limits and market pacing - the article reveals that
vertical scalability currently appears feasible in roadmaps upto about 100 cell
But the rate of 2D shrinks in successive 3D designs
will slow down from the recent historic average of 20% per generation to 5% -
due to the problems of registration which accumulate up as you add more
new technology report - How 3D NAND flash Stacks Up
January 15, 2014 - "In the 2D planar era, the basic underlying floating
gate technology (with a few exceptions) was essentially the same amongst all the
NAND flash manufacturers, however in the 3D era (which has recently begun) all
NAND flash memory
manufacturers are developing different 3D architectures" said Gregory Wong, President, Forward Insights
in a recent email introducing a new market report ($5,499) called
NAND Stacks Up (outline pdf) - which is co-authored with
NaMLab (Nano-electronic Materials
Laboratory) - in Dresden, Germany.
The new report describes the
various different approaches to 3D NAND design and provides an independent view
of the technical challenges which memory vendors have to solve to deliver
viable competing memories at different geometries.
Half Micron's nand flash now used in SSDs
January 7, 2014 - In a conference call related to financial results
for the quarter ended November 28, 2013 - which headlined on improved DRAM
results - Micron
- its nand flash business surpassed $1 billion revenue for the 1st time
- SSDs accounted for 48% of trade volume in nand flash (of which
2/3 was consumer
- in addition to traditional demand from the mobile market (phones etc)-
the company had identified
embedded applications in automotive markets as a business opportunity which
itself was taking around 10% of flash volume
- the big volume ramp for 3d nand flash was anticipated to be in the 2nd
half of 2015
Crocus petitions for dismissal of core STT patents
October 30, 2013 - Within the
SSD market all
those other types of of non volatile memory appear as mere driblets
compared to a sea of flash
memory - but that could change one day so it's worthwhile cementing sound
"We already done that thing" - aka "prior art"
- is the root of a petition (announced
yesterday) by Crocus
Technology for the US Patent and
Trademark Office to dismiss patent
(high speed low power magnetic memory device stuff) which is part of the IP
potfolio of competitor - Spin
by 2017 most flash will be 3D - says iSuppli
October 4, 2013 - In a
forecast yesterday IHS
iSuppli said - "by 2017 65% of all NAND flash memory chips
shipped worldwide will be produced using 3-D manufacturing processes, up from
less than 1% this year."
Editor's comments:- the
transition towards a new way of making flash memory (by vertical stacking of
deposition layers at the chip level) currently looks like a more viable way of
increasing flash densities in the long term - compared to shrinking the geometry
of cells - which is already straining the ingenuity of circuit designers to
counteract and manage
the impact of intrinsic defects in the materials which become more significant
as the stored charge for each virtual data bit gets smaller.
aspects of this trend toward shrinking 2D (aka planar) geometry - at the SSD
level - manifest as worsening raw metrics such as -
also:- market research
trust SSD market data?
Samsung offers 1st generation 3D nand flash SSDs for enterprise
August 13, 2013 -
it has started production of 2.5" SATA SSDs aimed at the enterprise market
- which use the company's new
3D Vertical NAND flash memories. Samsung says its 3D flash is
intrinsically more reliable, faster and uses less power than traditional 2D
flash at the same (10nm class) line geometries.
comments:- As SSDs - and compared spec by spec to any other SSDs - the new
V-NAND SSDs aren't remarkable - 960GB capacity and 35K
- which is what the market (in this case -
cloud storage array
But Samsung's new V-NAND SSDs are simply the first step
in the journey towards characterizing this new technology and to achieve
Samsung says its 3D technology could deliver
upto 24 cell layers vertically, using special etching technology that connects
the layers electronically by punching holes from the highest layer to the
When that happens - each wafer will be able to deliver an
order of magnitude more storage capacity from the same number of wafer starts -
using the same line resolution as traditional (planar) flash cells. (If you
think about the difference it made when the market went from SLC to MLC and then
again to TLC - the eventual market impact will be bigger than all those
combined.) But getting the chips and production equipment proven and economic
for double digit 3D cells will take years from where we are now.
each vertical layer takes additional processing time. In some ways it's like
adding more layers to your pizza - except that - the successive layers of
topping have to match up very precisely. (Around 2,000x more precisely than the
state of the art in metal additive technology - to give you an idea of the
difficulty and the elapsed time element.)
Crossbar has silicon for 3D RRAM
Editor:- August 5,
2013 - Crossbar today emerged
from stealth by
a working silicon demonstration of its 3D stacking technology which the company
says will enable the commercial use of RRAM in much higher capacity drives
Micron samples 16nm nand
Editor:- July 16, 2013 -
it will be in full production of 16nm nand flash (128Gb MLC memory devices)
in Q4 this year - and is designing SSDs around this process geometry - to ship
Crocus gets funding for x8 multibit magnetic semiconductor memory
April 8, 2013 -
announced it has
been awarded a contract from IARPA
to develop an 8-bit per cell memory based on its Magnetic Logic Unit
This will greatly reduce the energy consumed per
written-bit compared to any other memory technology, including DRAM, Flash,
SRAM and MRAM.
Lee, VP, product development at Crocus compared the 8 bits per cell
which the company thinks it can get from its MLU technology with the
state-of-the-art in nand flash - which is 3-4 bits per cell and also compared
to alternative magnetic semiconductor technologies like MRAM - which is
still only 1 bit per cell storage (SLC).
2017 could be 1st billion dollar year for non-flash nvm
February 18, 2013 - Yole
a new market report - Emerging Non-Volatile Memories (5,990 euros) which
describes why and how emerging alternative
NVM (FRAM, MRAM/STTMRAM,
PCM, RRAM) could grow from $209 million revenue in 2012 to $2 billion in
Among other things - the report says 3D RRAM could start to be
used in SSDs in 2017-2018, when 3D NAND's scalability prospects are
anticipated to worsen.
Everspin quadruples MRAM chip R/W
26, 2013 - Everspin
it will sample the first of a new family of MRAM chips in Q2.
MR10Q010 (1Mb in a 16 pin SOIC) has a quad SPI serial interface instead of the
single line interface offered in earlier MRAM devices. This makes it more
attractive for applications which need the simplicity of no wear-out
non volatile memory and
fast write performance in low capacity and small footprint applications.
Proton gets funds to rejuvenate flash
February 7, 2013 - Proton
Digital Systems today announced
the completion of its $2 million seed round to support continued development
and expansion of its LDPC-based flash read channel IP products that increase
and longevity of
Proton's IP is currently licensed for enterprise and consumer applications and
has already been adopted by some of the world's largest flash memory companies.
and DSP IP in SSDs,
how to market flash
management care schemes for SSDs
1/3 of Micron's nand flash trade sales go into SSDs
December 20, 2012 - Micron
that revenues from sales of its NAND flash products were 4% lower in the
quarter ended November 29, 2012 than they had been a year ago.
volume of the company's nand flash decreased by 9% - but average selling
prices increased 5%. Overall Micrion reported a net loss in the quarter of $275
million on sales of $1.8 billion.
call Micron said that SSD shipments had grown 20% compared to the previous
quarter. SSDs are 17% of Micron's nand business and the company estimates that
35% of the nand flash it supplies to trade customers end up in SSDs. MLC
was about 80% to 85% of nand flash wafer production with SLC and TLC making up
experimental technique eliminates flash endurance limit
December 2, 2012 - An article in IEEE
Memory Survives 100 Million Cycles - summarizes a recent research paper by
which described an experimental technique to redesign flash cells to improve
technique - which StorageSearch.com does not think is feasible to scale for
commercially competitive memory densities - involves designing addressable
heaters in the memory array which can pulse upto 800 degrees C for a few
milliseconds. This thermal "refreshing cycle" anneals the chip
material and heals common wear-out defects while also enabling the cells to be
"Afterward, we realized that there was no new physics
principle invented here, and we could have done this 10 years ago" said
the project director at Macronix
Micron in volume production of 1Gb PCM
18, 2012 - Micron
it was the 1st company to be in volume production of Phase Change Memory
The company's 45nm memories have upto 1Gb in a multichip
Editor's comments:- PCM fans will get excited about
But before we get carried away on a tidal wave of PCM SSD
speculation let's recall the reason we still use flash to implement the bulk
storage capacity in nearly all SSDs (despite flash's many
can be viable as an alternative to battery backed
RAM in the
cache part of a
flash SSD. Some SSD oems have already done that. But PCM's storage density
is too low to replace flash in mainstream SSD applications for at least the next
You can read more about various nvm technologies which were
going to make flash obsolete (including details of the 1st PCM PCIe SSD which
was unveiled a year ago) in my article
flash SSD's past phantom
Hynix acquires DSP SSD IP company LAMD
20, 2012 -SK Hynix
it has entered into an agreement to acquire California-based storage solution
The reason for the acquisition should be clear if you
read the article on my home page yesterday about the new generations of
adaptive SSD controllers. The
roadmap for flash memory
is dependent on these technologies to enable workable SSDs.
TrendFocus launches new memories in SSDs report
May 2, 2012 - have you ever wondered what percentage of a memory maker's SSD
output is SLC or MLC or TLC? and other things like that?
has launched a new NAND/SSD Information Service which includes that kind of
data. The company says that the SSD section of the report will include client
and enterprise SSD memory shipments and forecasts.
new nv market size report from Web-Feet
April 13, 2012 - Web-Feet
Research has consolidated the reported shipments from 17 main flash
makers to conclude that the flash memory market in 2011 was worth nearly $29
billion - an increase of 8% from 2010.
report - the 2011 Non Volatile Memory Market Shares by Vendor report ($2,500)
includes market shares by vendor for total nv memory (all types) and includes
breakdowns by vendor and forecasts. For more info contact:- Alan Niebel -
STT secures $36 million A round for OST-MRAM
February 15, 2012 - Spin
Transfer Technologies today
it has secured $36 million in Series A funding - led by its parent company,
Allied Minds and
Invesco Asset Management -
to accelerate development of STT's patented orthogonal spin transfer
magneto resistive random access memory technology (OST-MRAM).
"the company is poised to create the next generation of memory
applications combining the non-volatility of flash with the read and write
performance of DRAM and SRAM into one, seamless product."
Rambus gets into the nv memory IP market
February 6, 2012 -Rambus
it has acquired Unity
Semiconductor for an aggregate of $35 million in cash.
part of this acquisition, the Unity team members have joined Rambus to continue
developing innovations and solutions for next-generation
world's first PCIe PCM SSD
Editor:- June 14, 2011 -
NVSL ( the Non-Volatile Systems
Lab at UCSD) recently
a prototype PCIe PCM (phase-change memory) SSD - with R/W speeds upto 1.1GB/s
and 327MB/s respectively and 8GB usable capacity.
A spokesperson for
the Moneta SSD design team - Professor
Steven Swanson said "...Moneta gives us a window into the future of
what computer storage systems are going to look like, and gives us the
opportunity now to rethink how we design computer systems in response."
Swanson says he hopes to build the 2nd generation of the Moneta
storage device in the next 6 to 9 months and says the technology could be ready
for market in just a few years as the underlying phase-change memory technology
Editor's comments:- in a white paper
PCM Storage Array (pdf) the team outlines the design and architecture of
their PCM SSD prototype and also compares aspects of performance with entry
level PCIe flash SSDs from
Fusion-io. In a
I warned that you should not pay too much heed to comparative PCIe SSD
benchmarks - because from different arbitrary selected angles they can "prove"
different arbitrary performance rankings. I wouldn't be surprised if some
investors take fright that a PCM SSD scored higher than a Fusion-io SSD in some
of the published graphs. But for those who understand SSD architecture it
doesn't reveal anything new.
In my view this prototype clearly
demonstrates the strengths and weaknesses of PCM as an SSD technology.
PCM SSD strengths vs flash
The granularity of writes
in PCM is smaller and faster - which means that small R/W operations have higher
IOPS. If you have apps where that is important you can simply buy
SSDs with various
ratios of integrated RAM cache. That would give you small block IOPS
better than PCM - end of story. PCM has higher
than SLC - which means that the
overhead applied to endurance can be lighter than in most flash systems. Hence
potentially faster latency through to the media.
weaknesses vs flash
The prototype PCIe SSD card provides capacity
which is similar to RAM SSD
density - but with a large block R/W throughput which is much lower than
flash arrays. This
implementation used 16MB PCM chips.
Flash allows higher capacity writes
to a single chip - and this gives better peak performance results than PCM when
exploited in parallel architecture arrays. You can't get those flash peak
performance numbers from a PCM array in the same board footrpint - because many
PCM chips have to be written to concurrently to achieve the same capacity R/W
as a single flash chip. That means with today's technologies - flash SSDs
have a higher proportion of ready to write memory chips in the same chip count
population as PCM SSDs.
For more about alternative SSD technologies -
see SSD's past phantoms.
new report looks at NAND flash succession
January 11, 2011 - Forward Insights
and its research collaborators have compiled an in-depth, independent analysis
which analyzes the options for various
non volatile memory
technologies which could become viable in storage after floating gate NAND flash
hits fundamental scaling limitations
NAND? (pdf outline) is the product of experts in floating gate and charge
trap flash, and resistive and emerging memory technologies. This new report
(price $10k) evaluates 3D NAND and cross point memory concepts from Hynix,
Intel, Macronix, Micron, Samsung, SanDisk, Toshiba and Unity and concludes with
a roadmap till the end of the decade.
Toshiba integrates ECC into raw flash
April 5, 2011 - Toshiba
announced it is sampling
- 24nm flash memory chips (with upto 64GB capacity) with integrated ECC
controllers to simplify the design of consumer products which need storage.
new SmartNAND will provide our customers a smoother design experience into 24nm
generation and beyond," said Scott Nelson, VP, Memory Business
Unit, Toshiba America Electronic Components, Inc. "By enabling the system
designer to directly manage the NAND using a standard or custom host NAND
controller, while leaving the function of error correction within the NAND
package, SmartNAND results in faster time to market, access to leading
geometries and potentially lowers design costs when compared to conventional
NAND flash implementations with external ECC."
will Micron's enhanced flash really eliminate error concerns?
December 3, 2010 - Micron
recently announced availability of enhanced 16GB to 64GB 25nm
flash memory chips with integrated error management - which the company
says - removes the burden of ECC from the host and simplifies the use of flash
in enterpise apps.
Editor's comments:- as discussed in my recent article -
management in flash SSDs good blocks and less good blocks have always
coexisted in flash memory. But as device geometries shrink (to increase
capacity and speed) the margin of error between usable and non usable cells has
shrunk too. In practical terms this means that the raw media quaility of new
flash chips has declined in the past decade from under 1% defects, then 2%, 5%
and I've seen projections as high as 10% for emerging MLC.
read longer version of
new book - Inside NAND Flash
Editor:- November 17,
2010 - Forward
Insights (an SSD
analyst company) is one of the contributers to a new book called -
NAND Flash Memories.
The publishers say that
SSD designers must
understand flash technology in order to exploit its benefits and countermeasure
its weaknesses. The new book is a comprehensive guide to the NAND world -
from circuits design (analog and digital) to
new Samsung phone flash
Editor:- September 7, 2010 -
today introduced high-performance 16GB
embedded memory chips for use in smartphones.
A new feature enables the
host to interrupt a previously written write so as to respond sooner to
a higher priority read.
SSD readers intro to Nanocrystal Memories
June 30, 2010 - a recent blog from
of nanocrystal memories - a flash-like
nv memory technology from
table on Freescale's web site suggests 10x faster write cycle - and
(10 million cycles) than traditional flash. The technology is shipping in some
Macronix research pushes flash density
16, 2010 - Macronix
its research results related to its patented BE SONOS (barrier engineering)
charge-trapping technology which could make terabit NAND flash feasible.
Using 3D stacking, NAND Flash may achieve higher data storage capacity
and effectively lower fabrication cost without relying on advances in
lithography technology. Consequently some memory manufacturers have invested in
3D research recently.
Samsung ships 512Mb PRAM
Editor:- April 28, 2010 - Samsung Electronics
shipments of a 512Mb PRAM MCP which is is backwards compatible with 40nm
NOR flash memory in both
its hardware and software functionality allowing mobile handset designers the
convenience of retrofitting the 3x faster writing PRAM into exisiting
designs based on NOR.
Everspin samples industry's first 16Mb MRAM
April 19, 2010 - Everspin
it is sampling a 16Mb MRAM.
The MR4A16B is a 3.3-volt, parallel I/O
non-volatile RAM that features 35ns access times with unlimited read/write
cycles. Data is always non-volatile after each write for more than 20 years. In
addition, MRAM is immune to soft error rates associated with cosmic rays that
impact other memories. The 16Mb MRAM is organized as 1,048,576 words of 16 bits.
Pin and function-compatible with asynchronous SRAM, the MR4A16B targets
industrial automation, robotics, network and data storage, multi-function
printers and a host of other systems traditionally limited to SRAM-based
|"If these new memories
really are as good as the claims, why are we not seeing them in production
applications today? The answer appears to be inertia" - says
Brian Bailey, Technology
Editor - Semiconductor Engineering|
CEO of Intermolecular expands on
this by saying - NAND flash is a $30B industry that has tens of billions of
dollars in capital infrastructure that would need to be retooled. The big 4
players represent 95% of the market and they have a lot of existing investment.
The entire cost equation is CapEx, so they need to milk the tail of the revenues
as long as they can."
| Big Memory Shift
Ahead (May 2014)|
significant change over the past 10 years is the growth of NAND flash wafer
NAND wasn't even reported as a discrete category by the
SIA in 2000.
In 2005, all
NAND products only required 4.4% of total production wafers...
this had grown to almost 18% in 2012."
Managing Director |
Research (October 2012)
|In the quarter ended
November 28, 2013 - Micron said that about 50% of its nand flash memory bits
went into SSDs. |
|the Top SSD Companies -
Q4 2013 |
February 4, 2014
|10+ years of "MRAM
will soon replace flash"|
emerging size of
the flash SSD market as you see it today was by no means inevitable. |
owes a lot to 3 competing storage media competitors which failed to evolve fast
enough in the Darwinian jungle of the storage market.
||The article -
SSD's past phantom
demons explores the latent market threats which hovered around the flash SSD
market in the past 10 years. They seemed real and solid enough at the time.|
sudden power loss|
|Why should you care
what happens in an SSD when the power goes down? |
This important design
feature - which barely rates a mention in most SSD datasheets and press releases
- has a strong impact on
SSD data integrity
This article which reviews the architecture of power
line disturbance data integrity mitigation schemes in every major type of SSD
will help you understand why some SSDs which (work perfectly well in one type
of application) might fail in others... even when the changes in the
operational environment appear to be negligible.