Crucial and Samsung's V-NAND process node ?

[X.LINK]

Hi all,

Since 3D TLC SSDs came on the scene, manufacturers tend to avoid telling which process node they're using.
The only thing i've see, is that Samsung's 2nd generation V-NAND do use a 40nm process node.
And... that's it, nothing else since; apart from telling how many layers they do have. But that's not as useful as process node... or even a bitline or wordline size of the cell.

So what about Samsung's 3rd and 4th generation of V-NAND ? Is it still 40nm ? What about Crucial's V-NAND then ?

 

[Glaring_Mistake]

Regarding Samsung I would guess that they're still using the same (or similar) lithography as before though I can't be entirely certain of that.

On Crucial (or rather Intel/Micron) we happened to get into the topic of their choice of lithography in another thread not too long ago (starts at comment 13) : http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=thread...-binned-at-the-various-manufacturers.2510437/
Anyway they seem to be using both 16 and 20nm lithographies for their (Gen 1) 3D MLC NAND and the (Gen 1) 3D TLC NAND may likely also be using the same node(s).

 

[X.LINK]

That's some interesting lecture you got, thanks !

I don't know if IMFT still have 40nm fabs, but since going V-NAND allows them to reuse tools that made planar lithography, what you're saying is quite possible about 16nm and 20nm.

By the way, does using these two processes nodes at the same time means that the cell's size is 16nm x 20nm ?

 

[javier_machuk]

if micron is really using 16nm TLC for their v-nand that could be a problem... samsung 840 EVO long term data loss rings any bells?

I thought that one of the key benefits with v-nand is that they could use larger nodes in order to "reset" or reassure the reliability that was lost with the smallest planar TLC nodes.

Can somebody confirm this? I think that give samsung TLC v-nand an advantage reliability-wise, at least in my opinion.

 

[X.LINK]

It seems they'll never learn that such short terms decisions won't make them money in the long run...

So yes, V-NAND was also supposed to use larger nodes to have a decent reliability back, but it seems Micron isn't playing on the high-end segment anymore.
Since Crucial's MX300, the MX series tends to become closer to the low-end BX series than what the real MX series was.
I can't even imagine what will BX series become... QLC maybe ?

Also, since V-NAND was taped out, every manufacturers are disclosing any intel about the cells' process nodes or even anything about bitline and wordline sizes. I suspect them to slowly drift to smaller nodes, without us knowing it; so they can keep higher margins at the expense of our datas... and their reputation on the long run.

Kudos for having a confirmation about the process nodes, I still can't decide between buying a Samsung 850 EVO and a Crucial MX300.
Despite a suspected smaller node, this latter is still on the run with me since Crucial kept the Partial Power Loss Protection that no one except Intel SSDs also have, on the consumer market.

 

[Glaring_Mistake]

That's some interesting lecture you got, thanks !

I don't know if IMFT still have 40nm fabs, but since going V-NAND allows them to reuse tools that made planar lithography, what you're saying is quite possible about 16nm and 20nm.

By the way, does using these two processes nodes at the same time means that the cell's size is 16nm x 20nm ?

Well, either they have two litographies that are 20nm x xnm and 16nm x xnm or they have one lithography that is 16nm x 20nm but calls it 20nm for one drive and 16nm for another drive.
Both are possible but I would bet on the first one being true.

if micron is really using 16nm TLC for their v-nand that could be a problem... samsung 840 EVO long term data loss rings any bells?

I thought that one of the key benefits with v-nand is that they could use larger nodes in order to "reset" or reassure the reliability that was lost with the smallest panar nodes TLC.

Can somebody confirm this? I think that give samsung TLC v-nand an advantage reliability-wise, at least in my opinion.

Well, they've used 16nm 2D TLC NAND already in the BX200 which could suffer from dropping read speeds so it is not like that would necessarily prevent them.

That said, the MX300 using their 3D TLC NAND seems to be pretty resistant to voltage drift.
This is likely because of the controller because the primary reason read speeds dropped for the BX200 was the controller/firmware which means that it is possible that other drives using the same NAND may be more prone to voltage drift.
We'll see how the MX300 performs after some wear however since that can have quite an impact (to be clear I'm running tests on the MX300 and a number of drives to see how they handle voltage drift).

It seems they'll never learn that such short terms decisions won't make them money in the long run...

So yes, V-NAND was also supposed to use larger nodes to have a decent reliability back, but it seems Micron isn't playing on the high-end segment anymore.
Since Crucial's MX300, the MX series tends to become closer to the low-end BX series than what the real MX series was.
I can't even imagine what will BX series become... QLC maybe ?

As far as I know Crucial has not said what NAND the BX300 is going to use beyond 3D NAND but I don't think that they're prepared to start using 3D QLC NAND this early (or in this drive).
Have read that they do intend to release their 3D QLC NAND with 64 layers (second generation) and Intel 545s is already using 64 layer 3D TLC NAND so it should be on the way.
As for the controller Tomshardware believes that they will use the Silicon Motion SM2258XT (the XT at the end means it is a DRAMless version of the SM2258).