Lower Endurance—Why?

Below we have a diagram of a MOSFET (Metal Oxide Semiconductor Field Effect Transistor). When programming a cell, voltage is placed on the control gate, which forms an electric field that allows electrons to tunnel through the silicon oxide barrier to the floating gate. Once the tunneling process is complete, voltage to the control gate is dropped back to 0V and the silicon oxide acts as an insulator. Erasing a cell is done in a similar way but this time the voltage is placed on the silicon substrate (P-well in the picture), which again creates an electric field that allows the electrons to tunnel through the silicon oxide.

While the MOSFET is exactly the same for SLC, MLC and TLC, the difference lies in how the cell is programmed. With SLC, the cell is either programmed or it's not because it can only be "0" or "1". As MLC stores two bits in one cell, its value can either be "00", "01", "10" or "11", which means there are four different voltage states. TLC ups the voltage states to eight as there are eight different combinations of "0" and "1" when grouped in groups of three bits. Below are diagrams showing the graphical version of the voltage states:

SLC

MLC

TLC

The above diagrams show the voltages for brand new NAND—everything looks nice and neat and the only difference is that TLC has more states. However, the tunneling process that happens every time the cell is programmed or erased wears the silicon oxide out. The actual oxide is only about 10nm thick and it gets thinner every time a smaller process node is introduced, which is why endurance gets worse as we move to smaller nodes. When the silicon dioxide wears out, atomic bonds break and some electrons may get trapped inside the oxide during the tunneling process. That builds up negative charge in the silicon oxide, which in turn negates some of the control gate voltage when the cell is programmed.

The wear results in longer erase times because higher voltages need to be applied for longer times before the right voltage is found. Remember, the controller can't adjust to changes in program and erase voltages (well, some can; more on this on the next page) that come from the trapped electrons, cell leakage, and other sources. If the voltage that's supposed to work doesn't, the controller has to basically go on guess basis and simply try different voltages before the right one is found. That takes time and causes even more stress on the silicon oxide.

The difference between SLC, MLC, and TLC is pretty simple: SLC has the fewest voltage states and hence it can tolerate bigger changes in voltages. With TLC, there are eight different states and hence a lot less voltage room to play with. While the exact voltages used are unknown, you basically have to divide the same voltage into eight sections instead of four or two like the graphs above show, which means the voltages don't have room to change as much. The reason why a NAND block has to be retired is that erasing it starts to take too long, which impacts performance (and eventually a NAND block simply becomes nonfunctional, e.g. the voltage states for 010 and 011 begin to overlap).

There is also more and more ECC needed as the NAND wears out because the possibility for errors is greater. With TLC, that's once again a bigger problem because there are three bits to correct instead of one or two. While today's ECC engines are fairly powerful, at some point it will be easier to just retire the block than to keep correcting errors.

A TLC Refresher Lower Endurance: Hardly an Issue
POST A COMMENT

86 Comments

View All Comments

  • mayankleoboy1 - Monday, October 08, 2012 - link

    Samsung is known to release only a few firmware updates (unlike SF).
    But due to the somewhat quirky nature of TLC NAND, do you expect Samsung to release a newer firmware , with maybe better read performance , or better TRIM support ?
    Reply
  • Taft12 - Monday, October 08, 2012 - link

    Who cares?? Frequent firmware updates are a sign of an incompetent engineering and testing. I'll stick with the vendors known for getting it right the first time thank you very much. Reply
  • JuneBugKiller - Monday, October 08, 2012 - link

    If Samsung 840 120GB cost's $109.99 and you can get an OCZ Vertex 4 128GB for $104.99, which is faster in 80% of the benchmarks and is $5 cheaper then who's going to want it? So what if the price goes down to $80, are people going to save $25 thinking their hard drive is going to die twice as fast. I have bought (with works money) around 30 ssd's including Intel 80GB & 160GB Silver Case, Intel 710 100GB, Vertex 3 Max IOPS 120GB and Vertex 4 128. So far I've only had 2 ssd's go bad and they were both Intel 80GB ssd's. One wouldn't power on and the other reported at 8MB. Intel replaced them under warranty.

    The point is I have been buying SSD's for years and I just don't see how anyone would want these samsung ssd's. Samsung is famous for huge margins on each product. When a company pockets $240 off of a $500 tablet and their name isn't Apple then something is wrong.
    Reply
  • JarredWalton - Monday, October 08, 2012 - link

    Wow, talk about skewing the data by cherry picking the worst-case numbers for the 840. The Vertex 4 is only 85% faster in one specific test: AS-SSD Write performance. Of course, in the AS-SSD Read performance it's also 10% faster than the Vertex 4 256GB, but you just ignore that? Samsung also has some of the best SSDs in terms of large reliability figures, so even if the 840 is slightly slower than other drives in some tests, it may be the better option. Also don't forget to factor in that the 840 appears to be well-tuned for light workloads (e.g it's near the top of our light workload results).

    Personally, I think the 840 needs to come in below the current 830 drive prices to make sense, and it probably will not long after the official release. 128GB 830 drives already go for under $100, and 256GB drives have been at $200 for over a month now -- likely all in preparation for the release of the 840. TLC NAND is cheaper to manufacture (per GB), and long-term it will be significantly more profitable for Samsung. Get some good DSPs added into the mix and I wouldn't be surprised to see most SSDs in two generations being TLC based, with MLC moving to the enterprise level and SLC basically going away because it's too expensive.
    Reply
  • JuneBugKiller - Monday, October 08, 2012 - link

    No I'm talking about counting each benchmark add the total and divide the number by how many OCZ Vertex 4 won and it was over 80% of the total number of benchmarks. How is Samsung a better option? TLC over MLC, OCZ fastest drive to Samsungs slowest new drive. Of course Samsung is going to make more profit but why would you want to spend the same amount on a slower drive with less endurance? Reply
  • JarredWalton - Monday, October 08, 2012 - link

    Sorry, 80% of benchmarks is correct; I read that wrong. But let's put that in perspective:
    http://www.anandtech.com/bench/Product/678?vs=628

    If we look at average performance across all benchmarks, the difference between OCZ Vertex 256GB and Samsung 840 250GB is a 12.4% advantage for OCZ. However, OCZ hasn't exactly been free from firmware issues. That right there is the reason many people will pay a bit more for a Samsung (even if it's slower).

    Would I buy an 840 right now for $200 or whatever? Definitely not -- I'd actually take the 830, just for proven reliability over time. Give the 840 a couple months just to be safe, then check the prices. If it's still more expensive than the Vertex 4, sure, go for OCZ if you'd like. If they're the same price, though, the 12% performance is practically meaningless for most consumer workloads.
    Reply
  • sean.crees - Monday, October 08, 2012 - link

    Because it's not just about performance, it's also about reliability. This is user data we are talking about. Even a single loss could be catastrophic. OCZ doesn't have the reliability track record that Samsung and Intel has, and for that reason there are many people who will only ever buy from Samsung or Intel. So then OCZ isn't even mentionable. It doesn't matter if they are faster because who cares if your data is at risk?

    Also you have to consider that the REAL WORLD difference between any modern 6Gbps SSD is negligible, so then performance means even less. It ends up being a contest of reliability instead of a contest of speed. In that contest, OCZ loses.
    Reply
  • krumme - Monday, October 08, 2012 - link

    I would want it :)

    We have about 7-8 ssd in the house, about 3 of those is vertex 3 ssd, and there have been realiability and firmware issues.

    I think ssd speeds since the last 2 years have been good enough and would prioritize reliability any day.

    I had to replace an sandisk u100 in my Samsung 9 series x3c to a faster one, but that was because the u100 was like a return to 4 years ago. Its a good backup now.

    Now hopefully reliability is there, and prices will go down so everyone can afford it. We dont need more firmware updates and shit. Its like the first 3d gfx in the mid 90, - a mess.

    But you are right Samsung is starting to get expensive, and charge for the brand. Wether you like it or not it will certainly mean more Samsung reviews too. Being a big boy, have advantages :)
    Reply
  • Taft12 - Monday, October 08, 2012 - link

    "If Samsung 840 120GB cost's $109.99 and you can get an OCZ Vertex 4 128GB for $104.99, which is faster in 80% of the benchmarks and is $5 cheaper then who's going to want it?"

    A person who values reliability at a value higher than $0 will choose the non-OCZ product every time.
    Reply
  • Jaybus - Thursday, October 11, 2012 - link

    Time will tell. The TLC NAND should end up around 30% cheaper per GB than MLC. Then it will come down to buying a 240 GB MLC for $200 or a 320 GB TLC for the same price. Reply

Log in

Don't have an account? Sign up now