Groups | Search | Server Info | Login | Register

Groups > alt.comp.hardware > #20732

Re: How Do SSDs Wear Out?

Show all headers | View raw

On 2025-02-18 06:28, Paul wrote:
> On Mon, 2/17/2025 3:46 PM, Carlos E.R. wrote:
>> On 2025-02-14 09:11, Paul wrote:
>>> On Fri, 2/14/2025 12:47 AM, Boris wrote:
>>
>> ...
>>
>>> The physical cells, the structure at the atomic level, is
>>> damaged by the writes.
>>>
>>> Each cell has a "voltage" stored on it. Established by putting
>>> some electrons on a floating gate. The path for this is
>>> quantum mechanically disallowed, and to get the electrons
>>> onto the gate requires tunneling. The electrons will sit
>>> on the gate for up to ten years (retention time estimate, info
>>> on this has not been updated in a long long time so we are left
>>> to guess whether it scales in any way with gate size).
>>
>> I wonder if we can store the disk for five years, then plug it in and somehow refresh the charges in the cells.
>>
>> ...
>>
>>> By mapping the sectors, using a mapping table, and "moving the MBR around
>>> each time it is written", that is wear leveling. The drive has a pool of
>>> unwritten blocks. On a write request, an unused block is written.
>>> Perhaps the block is at address 27, and it contained MBR sector 0.
>>> The map file the drive keeps then, it has to remember that aspect.
>>> On a read, we request sector 0, the map goes "oh, that is block 27",
>>> and the drive does the read at that address, and there is our MBR.
>>> Now, if I abuse the MBR by writing it a lot, a hole isn't burned in it.
>>> The sector has been "virtualized", and only the mapping table knows
>>> where my sector is stored :-)
>>
>> Where is the map stored? I always wondered about this.
>>
>> ...
>>
>>
>> Thanks a lot for the summary :-)
> 
> The issue of "recharging cells" has already come up, with respect
> to TLC. For at least Samsung, they may have a provision for doing that.
> 
> SSDs don't have a real time clock, so they cannot tell five years has
> passed. All they have is power-on-hours, which is a useful metric for
> an SSD that is alive and working every single day.
> 
> The drive can tell when a sector is getting "spongy" due to the
> error count. A TLC sector, might have a bit in error, for every sector.
> Correcting all the sectors is nothing new for the drive. And since
> the syndrome is 50 bytes for a 512 byte sector, that's a *huge* syndrome
> allowing a lot of bits in error to be corrected. The drive can allow
> the TLC cell to have more and more errors in it. Then, once a portion
> of the error capability is used up, the drive could re-write the sector.
> 
> That's one way they could do it.
> 
> But since no company has enthusiast promoters like in the OCZ days,
> we cannot get information from company reps about how things work.
> 
> *******
> 
> The map can be stored in an "SLC-like" critical data storage
> area of the flash. But that's not the part I am particularly
> interested in. I'm more curious about how a map file can be
> maintained, without burning a hole in the SSD while doing so.

Exactly!

> It is the handling policy of the map file, whether it is journaled
> or protected in some way, that I am curious about.
> 
> But don't expect to find an honest explainer page on the web.
> 
> All we know, is the power can go off, and the SSD drive seems to survive.

Ok :-)



-- 
Cheers, Carlos.

Back to alt.comp.hardware | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

How Do SSDs Wear Out? Boris <Boris@invalid.invalid> - 2025-02-14 05:47 +0000
  Re: How Do SSDs Wear Out? Paul <nospam@needed.invalid> - 2025-02-14 03:11 -0500
    Re: How Do SSDs Wear Out? "Carlos E.R." <robin_listas@es.invalid> - 2025-02-17 21:46 +0100
      Re: How Do SSDs Wear Out? Paul <nospam@needed.invalid> - 2025-02-18 00:28 -0500
        Re: How Do SSDs Wear Out? "Carlos E.R." <robin_listas@es.invalid> - 2025-02-18 14:26 +0100
    Re: How Do SSDs Wear Out? Boris <Boris@invalid.invalid> - 2025-02-18 01:42 +0000
      Re: How Do SSDs Wear Out? Paul <nospam@needed.invalid> - 2025-02-18 03:03 -0500

csiph-web