Overcoming Hurdles
A paper, titled “The Bleak Future of NAND Flash Memory” was recently jointly published by the University of California and Microsoft Research. It has been picked up by many media outlets who all seem to be beating the same morbid drum, spinning…
A paper, titled “The Bleak Future of NAND Flash Memory” was recently jointly published by the University of California and Microsoft Research. It has been picked up by many media outlets who all seem to be beating the same morbid drum, spinning tales of a seemingly apocalyptic end to the reign of flash-based storage devices. While I agree with some of what these authors have to say, I have reservations about the methods upon which the paper is based.
TLC and beyond?
The paper kicks off by declaring steep increases in latency and drops in lifetime associated with increases in bits-per-cell. While this is true, flash memory manufacturers are not making large pushes to increase bits-per-cell beyond the standard MLC (2 bits per cell) tech. Sure some have dabbled in 3-bit MLC, also called Triple Level Cell (TLC) which is a bit of a misnomer since storing three bits in a cell actually requires eight voltage level bands, not three as the name implies. Moving from SLC to MLC doubles density, but the diminishing returns increase sharply after that – MLC to TLC only increases capacity by a another 1.5x, but sees a 2-4x reduction in performance and endurance. In light of this, there is little demand for TLC flash, and where there is, it’s clear by the usage cases that it is not meant for anything beyond light usage. There’s nothing wrong with the paper going down this road, but the reality is that increasing bits per cell is not the envelope being pushed by the flash memory industry.
Wait a second – where is 25nm MLC?
Looking at the above we see a glaring omission – 25nm MLC flash, which has been around for close to two years now, and constitutes the majority of shipping flash memory parts currently in production. SLC was also omitted, but I can see the reason for this – it’s hard to get your hands on 25nm SLC these days. Why? Because MLC technology has been improved upon to the point where ‘enterprise MLC’ (eMLC) is rapidly replacing SLC even despite the supposed reduction in reliability and endurance over SLC. The reasons for this are simple, and are completely sidestepped or otherwise overlooked by the paper:
- SSD controllers employ write combination and wear leveling techniques.
- Some controllers even compress data on-the-fly as to further reduce writes and provisioning.
- Controller-level Error Correction (ECC) has improved dramatically with each process shrink.
- SSD controllers can be programmed to compensate for the drift of data stored in a cell (eMLC).
The paper continues with a hard push on a ‘TLC is bad’ mantra, and they even briefly mention four-bit-per-cell.
Wait, what’s a Constant Die Count?!?!
The paper then moves on to place a broad reaching artificial limit on the methods of testing and evaluating the future iterations of flash memory technology. They assume a Prototypical SSD with a ‘Constant Die Count’, presuming that all flash controller architectures moving forward will be forced to use a fixed number of flash memory dies. This is, simply put, an absurd assumption. tantamount to limiting yourself to a single core and then writing a paper about ‘The Bleak Future of CPU’s’. To understand what’s going on here, lets recap the current hurdles seen by the flash memory industry:
- As the process shrinks, program (write) time and latency increase.
- As the process shrinks, write cycles (endurance) decreases.
…and now for the ways around these issues:
- Increasing parallelism (die count) effectively counters latency, especially when Native Command Queueing (NCQ) is properly employed to allow that parallelism to be effective as workloads increase.
- Increasing parallelism increases throughput (simple math).
- Increasing die count increases the available capacity, which increases endurance given a constant write workload (the writes are distributed over a greater area, therefore each cell is erased less often).
Given the above, fixing die count shoots all of these no-brainer architectural improvements in the foot. This rings even truer when you consider how the cost of each chip (full of dies) will drop over the coming years. Don’t take my word for it – this pic is from the paper itself:
Even without increasing die count, each die shrink doubles capacity, so even with the Constant Die Count restriction, longevity is effectively maintained. Also, with these die shrinks have come other improvements to flash architecture – both to increase throughput and decrease latency. Another thing sidestepped by the paper is the difference in quality of flash memory. This varies widely among manufacturers, and was highlighted in a recent presentation at the Flash Memory Summit, where several different flash memory types were tested. While some flash actually performed below spec, others performed well above it.
2x nm MLC flash outperforming its rated spec.
If you flip through that presentation yourself, you will see an ironic twist: many of the larger feature size parts (3x nm / 5x nm) performed below their specification. This is because as manufacturers have shrunk the dies, they have simultaneously improved upon the process. Further, a given process node is improved upon as it is executed within its life cycle. For example, Intel 25nm flash did not appear in consumer and and finally enterprise devices for months after its manufacture begun. This is because while CPU bases its yield on more of a go / no-go test (i.e. it either works or it doesn’t), flash memory has an added metric of quality. As the plant fine tunes their production, not only do they get a higher yield, but they get higher enduring parts as well. Where CPU’s are binned according to their maximum speed, flash memory is binned according to its speed and its relative quality (endurance). IMFT announced 20nm flash last year, but it’s not in shipping products because the quality and endurance are not yet mature enough for mass production and use.
Conclusion
The point I want all of you to take home here is that just as with the CPU, RAM, or any other industry involving wafers and dies, the manufacturers will adapt and overcome to the hurdles they meet. There is always another way, and when the need arises, manufacturers will figure it out. The main drivers of this paper were TLC (and higher), and a CDC controller architecture. Nobody else really takes TLC seriously, especially for enterprise use, and no SSD manufacturer worth their salt is going to keep the die counts fixed. Both of these fallacies effectively negate the entire premise of the paper.
To be clear – I am not blind to the issues flash memory faces moving forward. That said, I can’t understand what the motives of this paper are, or why they overlooked important points and skewed their results so severely, but to me it is just bad science. If you want to make a point, don’t overreach to the point of undermining your message.
UCSD is looking to get
UCSD is looking to get attention and research dollars for alternative technologies like PCM and MRAM. This is the same group that built a working PCIe PCM memory board last year.
I believe the same reference architecture was used in the test bed that produced the data in the paper. Despite the hypberbole, TLC (wrong) terminolgy and ignoring solutions that are being developed, the paper did list the challenges that face NAND Flash producers as cell sizes shrink.
What they avoided discussing was the billions in development going into the solutions by the foundry owners that will keep flash viable for at least 20 years.
Thanks James. While writing
Thanks James. While writing this I kept hoping there was some underlying motive – otherwise it would have been just plain bad writing.
Having worked at Fairchild
Having worked at Fairchild and then Intel over a career of 35+ years I have observed the following facts of the semiconductor industry.
1.) Intel rules the roost; lock, stock, and barrel.
2.) Anytime an upstart (ie. MIPS, SGI, Sun, AMD, Weitek, Moto, PowerPC, ARM, etc.) makes a run for the lead, 2-4 years pass and Intel reaches into the magic bag and comes out with just enough technology put it down hard.
3.) Advancement in the evolution of semiconductor technology at Intel is dished out to maximize profits, period.
4.) Intel has more semiconductor patents and processes developed and in waiting than you can imagine. Moore’s law is alive and well for the long term foreseeable future.
5.) Expect cheap 1-5TB SSDs from Intel in 2013-2014.
As always, Allyn’s
As always, Allyn’s observations on IT storage technology must be given serious consideration.
It’s too bad that UCSD did not submit their paper to any kind of objective peer review: maybe they expected that to happen de facto with changes the Internet has brought?
As a 2-time graduate of the University of California, I can personally attest to the fact that its faculty and staff have fallen for the lure of national socialism (read “nazism”).
When you combine Charlotte Iserbyt’s observations with the trend to “commercialize” education, complete with new Federal laws preventing students from discharging student loans in bankruptcy court, is it any wonder that famous universities have now descended into greed?
To finish this story with a very painful and personal experience, UCI stole a book I wrote in 1992, modified it, and hosted it on the Internet along with 19 other major U.S. colleges and universities.
When I sued all of them, they all attempted to appear in Court with UNlicensed California attorneys. So, I won by default, per U.S. v. High Country Broadcasting.
So, on top of being consumed with greed, UC is now legally bankrupt:
http://www.supremelaw.org/copyrite/uci.edu/invoice7.htm
/s/ Paul A. Mitchell aka “MRFS”
http://www.supremelaw.org/cc/aol/
Allyn:
Excellent points on
Allyn:
Excellent points on the original paper. I’d love to see this as a PDF. Any chance you could publish in that format either here, or on your Google+ page?
Thanks.
Scott
Tips on how to Talk to an
Tips on how to Talk to an established medical-related Negligence Attorney at law With regard to Handbook with Shelley Casey
Frequently the first concern the moment phoning a law firm inquiries the charges contained. With lawyers having some many excessive fat for each and every hour that will or even testosterone ranges, likely to obvious can’t stand that your effective quick message cost your fortune. Why then you shouldn’t we start by getting considering the accurate secret issue:The total amount Should really This Charge To determine Some Health-related Disregard Experienced Rate?Found in at the moment, and having availability regarding law firms heightened if you utilize the web, the simple react to is that you could locate a Health care Negelect Attorney that are able to have a zero cost original mobile phone chat in addition to a initial impression.As soon as they currently have brought up ones own possible are saying on hand, utilised together also be effective at let you know out of the tips for assisting remarks meant for health-related forget about. This could and must begin to add some following alternatives:
Appropriate Aid
No Succeed Zero cost by using a Conditional Charge Settlement
paying just by instalments
Lawful AidLegal Handbook is quickly intended for professional medical neglectfulness claims to deal with a lot less than 16 plus in several other problems. Then again, is extremely important trying to get 100 % legal Device can postpone is significant as well as it may be much easier to abide by protection declare inside a Based Service charge Bargain.Dependant Cost AgreementsA Depending Monetary fee Recognizing, and No End up getting Cost-free package indicates you could do your personal health-related negelect state without paying a person’s ideal costs in the act early when using state. Your attorney believes to wait to get costs ’till the completion over the promise, to get being employed by free of charge assuming that hawaii will require you were eligible for yet another significance, called the successes transaction. Any true blessing yourself which you do not ever will need to purchase the particular legal representatives costs because of the scenario carries on, which may face thousands of unwanted weight every now and then.Forking out As a result of InstallmentsIf one doesn’t qualify for often while using the more than plans, you could be eligible for consider a tiny monthly payment while using the legal professional, or maybe anyone to deal with ones own price ranges on instalments. The remedy relies around the solicitor’s different ailments and suppleness as well as the strength of that scenarios keep.These StepOnce maybe you have supplied the main points of this lie state a person’s Medical related Negelect Attorney at law will need advise to look at the will probably be worth of this allege. You could prioritise opertation making use of the attorney, and typically now you might agree with the fact to transmit almost any data and ideas towards the legal representative by just document. Regardless of what way you have got, any lawyer is normally about to would like these particular knowledge (for it to be benefits putting together it again with regards to your animal before you start):
the evening out in the remedy and when you get one awakened into the point the very first thing was missing incorrect
the facts connected with a condition meant to the National medical product as well as the Doctor and even Dental office professionist
info on the injuries or simply pain and even suffering
information upon your loss and also bills
more details inside of your cash to make certain that all of the backing prospects may be considered
Paperwork To the Heath care treatment Mistake Solicitor You will attorney will surely have to be aware of the subsequent to:
clones of the healthiness data and even frequent records
a usage skin color appointments that you joined
what they can be termed within your medical professionals or common cosmetic surgeons who seem to given your treatment methods method
Discovering Any ClaimOnce any Medical-related Fail Authority has got repeating this pieces of information they can provide a good also analysis from your have. This particular can widely be a principal look at however since it is not even until eventually professional medical stats are picked up they will form one further research your announce. Even so, a attorney at law need to provide:
an understanding of the prospective buyers about positive results of the insist (when you)
an concept of the level of pay up if your claim is certainly productive
an idea of this reputable obligations linked with and create a health-related negelect state
the predicted time period that your choice of declare requires
How Are you presently Afflicted Future?Following any health issues mistake attorney at law provides the countless information and facts recommended to verify your state, you will leave it every towards your ex lover to look at up coming behavior. They will acquire well being maintenance data files, show physician and even perchance an important advice, in the event that a person’s maintain proceeds they may are given each one of the data files were required to rate your maintain.
Ultima scarpe Hogan in hogan sito ufficiale bugger off,Hogan sito ufficiale hogan outlet,Hogan interactive.Ultima collezione,Il massimo della qualit??,Spedizione gratuita,Fino al 70% di sconto.sconto Hogan Ballerine Donna,Hogan sito ufficiale,Scarpe Hogan vendita in hogan hogan sito ufficialemofette.nuovo Hogan Hi-Sprint Uomo hogan outlet,Hogan Shop,Hogan scarpe in vendita hogan sito ufficialein Hogan sito ufficiale.Scarpe Hogan Interactive Donna,Hogan Give someone the sack,Scarpe Hogan hogan outlet in vendita in Hogan sito ufficiale.
http://www.dongfeng8.com/home/space.php?uid=28800&do=blog&id=516811
http://resourcemedia.tv/node/538#comment-15660134
http://www.1919qq.com/bbs/viewthread.php?tid=2020774&pid=2234584&page=1&extra=page%3D1#pid2234584