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

  • iwod - Tuesday, October 09, 2012 - link

    I would be buying an Intel instead if it cost $109.99. While Sammy is possibly the closest to intel in terms of reliability. I wouldn't want a TLC SSD for the same price. And the difference would be marginal anyway. I could feel the difference between a decent 3Gbps SSD and 6Gbps SSD, i am not sure if there is anything to feel for two decent 6Gbps SSD.

    Now if the retail price is really in the range of $70 - 80 that would be a huge difference. Because you are essentially choosing the second best reliable SSD for decent performance and cheapest price.
    Reply
  • Death666Angel - Tuesday, October 09, 2012 - link

    The Samsung 830 is the 7th cheapest SSD (based on €/GB) I can find on my go-to price comparison website. 3 of the six before it are much slower Kingston/OCZ drives and the other 3 have SF controllers. Considering the kind of inconsistent performance that delivers and the track record, I would go with the Samsung all the time.

    Now, the 840 just launched and the pricing here is 25% above the 830. But it isn't even available for shipping anywhere. So I don't think that your comparison based on the Vertex4 ship has been shipping for months has any relevance. Also, aren't the Vertex4 numbers with their turbo-mode enabled which stops working after filling the drive to 50%? The Vertex4 256GB model costs 175€, the Samsung 830 256GB model costs 148€. Do you really think the 840 will be more expensive than the 830 in a couple of months? And even so, do you think the Vertex4 if that much better for consumers than the 830? I just don't see it.
    Reply
  • Blazorthon - Monday, October 29, 2012 - link

    Actually, the slowing down after 50% and such for Vertex 4 was fixed in firmware 1.5. The current numbers don't rely on Turbo, they're all the time.

    Also, to all of you, Vertex 4, unlike its predecessors, uses a Marvell-designed controller. Anyone who's used a Crucial or Plextor SSD probably knows that Marvell based SSDs are about as reliability as you can get, up there with Samsung and Intel. Vertex 4 is OCZ's most reliable SSD to date by huge margins, especially with current firmware where it rivals even Samsung and the others in reliability unlike the original firmware (which was specifically stated to be in beta, so anyone who bought Vertex 4 around launch time without doing their research and finding that out probably isn't someone who should be allowed to buy their own SSDs without expert help).

    Also, IDK about UK prices, but in the USA, the Vertex 4, Samsung 830, Samsung 840, and occasionally also the Crucial M4 and Plextor M5S are all at about the same price points for 64GB, 128GB, and 256GB capacities right now. As someone who has experience with Vertex 4, I'd definitely recommend it over Samsung 830 and Samsung 840 so long as it's not more expensive.
    Reply
  • Assimilator87 - Monday, October 08, 2012 - link

    How come the OCZ 4 series just suddenly disappear when the review gets to the power consumption tests? Reply
  • JarredWalton - Monday, October 08, 2012 - link

    Because Kristian imported the results from SSD Bench for the "1.4 Firmware" for the OCZ drives, and when those were retested Anand didn't add in power use figures. I've added them to the graphs using the original power figures at release. Thanks for the heads up! Reply
  • Coup27 - Monday, October 08, 2012 - link

    Excellent article Kristian, enjoyed it.

    Some constructive criticism, although I appreciate that English is not your native language. Sentences aren't that clear when you use a double negative, such as "I wouldn't find it unlikely...". It would be much clearer if you had wrote "It would be likely..."
    Reply
  • Kristian Vättö - Monday, October 08, 2012 - link

    Point taken, thanks for the feedback :-)

    Some things sound logical in my ears but they may not make much sense for a native speaker. I guess this is partially because in English exams and stuff, using more difficult sentence structures (like double negative) will yield higher scores, even though they are not commonly used by natives. Will try to pay more attention to this in the future.
    Reply
  • Sufo - Tuesday, October 09, 2012 - link

    I don't know if I agree with this. While double negatives are certainly inappropriate in some situations, the example you give is nit picking. Add to that, there is a subtle (albeit not technical) difference between "I wouldn't find it unlikely" and "I would find it likely". I accept that had Kristian made a legitimate grammatical error, then correcting it (while a little pedantic) may have been justified, but simply editing for style seems unnecessary.

    I guess my point is Kristian was not objectively wrong, or even unclear (although admittedly, complexity for complexity's sake benefits no one) and so he shouldn't pay too much heed to comments like this simply because OP is an English speaking native and he is not. I am, however, confident that he's more than capable of coming to his own conclusions on the matter :P
    Reply
  • smartthanyou - Monday, October 08, 2012 - link

    I am probably in the minority on this, but as a matter of policy, I don't think you should call anything a review without pricing and a approximate release date.

    Particularly the price. If I don't see a price, I don't bother to read the article.
    Reply
  • sean.crees - Monday, October 08, 2012 - link

    Price is on the 5th page labeled "The Samsung SSD 840" about half way down.

    Although i noticed the price for the 840 250GB is the same as the 840 Pro's 256GB? Is this a typo? Why would anyone buy the non pro with 6 less GB and slower speeds for the same price?
    Reply

Log in

Don't have an account? Sign up now