| leading the way to the
new storage frontier||...|
trust SSD market data?
sudden power loss
flash story... could it have been simplified?
|the unsung hero of 3D
|Editor:- September 15, 2015 - One of the
interesting surprises about 3D
nand flash which emerged
was that the endurance
- reported by SSD designers - was better than you would have expected if you
had taken as your starting point - assumptions about 2D nand with similar
capacity and next generation (smaller) planar geometries.|
Part of the
explanation - which I discussed last year - was the use of different materials.
But there's more to it than that as I realized today in a new (to
me) white paper
V-Nand paradigm shift (pdf) mentioned in an email blast from Samsung.
Samsung's version of 3D - the charge is held in a silicon nitride charge trap -
which has better insulating properties than conventional 2D materials - and
lower leakage. (Essential if you think about it to even get a hope of doing
more levels.) That's the bit I already wrote about.
What I didn't
appreciate before was this.
The SiN charge trap requires a lower
voltage to program each cell than traditional 2D (floating gate) designs. (The
destructive effect of the energy in these write pulses is one of the
contributory factors to cell wear in 2D.)
The combination of better
insulation and less destructive write process would give you a better endurance
figure for the same flash process geometry in nanometers.
But a 3rd
element in Samsung's 32 layer 3D (which you could say is the sneaky pragmatic
business decision bit) is that these devices weren't 2Xnm geomtery at all.
Samsung decided to bring this technology to market with a mature
proven 3Xnm process. (3Xnm gives you better endurance anyway than 2Xnm in 2D
So when Samsung talk about their 32 layer V-nand - with 35,000
P/E cycles "offering 10x increase in
over the 3,000 cycles provided by planar (2D) NAND" there are several
ingredients which combine to get this headline figure. And that's why this
paper is useful.
As to how much of that "10x" will continue
to be sustainable as more layers are added and as cell sizes get smaller - will
remain to be seen. Samsung's paper suggests that they are hopeful about aiming
at 100 layers in the next few years.
But it's clear that a lower
write pulse voltage is one of the unsung heroes in this better endurance
the white paper
PS - not everyone is convinced about the
market transforming imminence of Samsung's 3D nand.
In his August
2015 paper -
Technology:Annual Update (pdf) - Jim Handy founder Objective Analysis
said "Samsung is shipping at a loss and we have found no evidence of a
Furthermore Jim is sceptical about the 3D nand
industry's assertions that " the big ramp will be in 2017."
|3D nand fab
yield - the nth layer tax?|
|Editor:- July 5, 2017 - There has been much
uneasiness and disquiet recently in markets which use
nand flash memory chips
about the shortages - which are the most obvious sign that progress towards
successor 3D technologies has not been entirely plain sailing. |
much information and
many blogs in the market I think some fundamental indicators are not
publicly known and their contributions may only become understood after the
industry returns to business
If you have the time and inclination to read more about these problems
take a look my new blog -
are more dimensions of analysis needed to get a clearer picture of future 3D
|The first time I saw the
term "Garbage Collection" in an SSD context on StorageSearch.com was
in 2002. |
It was in an article about flash SSD reliability.
|flash SSD Jargon Explained|
|Even if you already knew
that adaptive R/W and DSP ECC 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 in 3D.
Among other things DensBits says that the scope for inter-cell
interference grows from 8 identifiable routes in 2D to 26 for each cell in 3D.
care management & DSP IP in SSDs
|SSDs are only as good as
the people who design them and make them. |
There can be orders of
magnitude difference in operational outcomes - even when different SSD makers
are using exactly the same memory chips.
|principles of bad
block management in flash SSDs|