free web stats

Race towards Bottom Continues with 16nm NAND

by: admin Thursday, November 29th, 2012

Considering today’s speed of technical developments, mainstream SSDs appear to be on track to become increasingly less durable until they’re barely suitable even for consumer laptops.

TLC NANDMost NAND Flash manufacturers are already down to 19 or 20 nm technology and they are reportedly eyeing increasingly smaller and cheaper ways of producing flash circuitry.

The advantage of 16 nm technology is that the modules are reduced further in size, some 30 % from the current 20 nm process. One of the many downsides is that the yields will deteriorate initially, which might actually serve to jack up prices due to low availability, as opposed to the other way around.

We are used to things getting cheaper as the production process drops, but apparently this might not happen, according to DigiTimes.┬áThis could be a good thing for the manufacturers, according to the reports, as it will “help the NAND flash industry return to a healthier supply-demand dynamic”, says “sources”.

In virtually all other areas of silicon manufacturing, such as CPUs and GPUs, a smaller production process is usually a step forward for several reasons, most of all efficiency.

However, this is not necessarily the case with the NAND flash modules found in solid state drives, as we have already established. The good part is that SSDs do indeed become more affordable, but ever smaller transistors have less space to hold a charge and wear down much, much faster. The same is rule applies to SLC vs. MLC nand, the latter of which are considerably less durable.

Now we have even started to see TLC NAND in products like the Samsung 840 (not the PRO model), and it will be very interesting to see how these drives stand the test of time. Some of these problems can be solved with better error correction and over-provisioning, but eventually a limit should be reached.

« | Home | »

3 Responses to “Race towards Bottom Continues with 16nm NAND”

Thomas Said:

You always have a negative view, can you not post something positive, possibly highlighting new storage technology which solves these problems…

Comment made on November 30th, 2012 at 10:39 pm
vt Said:

Something positive may come with a “new” storage technology, so far the manufacturers are just getting the most $$$ out of their “investments”. At the cost of end-user, of course. That’s going to happen till they are able to squeeze the last possible $ out of NAND by making pure-marketing claims about “advancements”. If the customers were even dumber than they are, the manufacturers would, for example, even make up some crap like “we decided to improve the NAND chips by putting pure gold into them” and force the customers pay 10x of that gold price based on the marketing deception.
In fact the argument like “One of the many downsides is that the yields will deteriorate initially, which might actually serve to jack up prices due to low availability, as opposed to the other way around.” coming from the manufacturers is a perfect example of marketing lies, when they argue that people have to pay more for worse quality NAND just because it is “new”.
They good thing is that they are approaching the limit when it is no longer possible to gain any more profit by spreading marketing deception about “new technological process” and such. The marketing bubble about “advanced NAND memory” so far remains intact only due to increased NAND marketing share in mobile devices sector.
Another positive thing is that in about 2 years due to the marketing bubble about NAND and potential instability of the artificially inflated demand and prices they may actually move to sampling actually more advanced devices based on ferroelectric and resistance memory.

Comment made on December 2nd, 2012 at 8:22 am
vt Said:
For some actual increase of the available NAND erase cycles, it may be a good choice compared with the current pulse erase configuration based life extension methods (coupled with minor silicone changes) for the so-called “enterprise NAND”. However, the overall error rate will slightly increase with every “heating cycle”.

Comment made on December 3rd, 2012 at 7:41 am

Leave a Comment