News Kioxia Says 3D XPoint ‘Isn’t the Future’

[Admin]

Kioxia has taken aims at storage-class memories like Intel and Micron's 3D XPoint and said it isn't the future, at the recent IEDM conference. Its argument is that it is costly to scale to many layers, unlike 3D NAND.

Kioxia Says 3D XPoint ‘Isn’t the Future’ : Read more

 

[bit_user]

First, I don't see why SCM couldn't also be scaled in the same way as 3D NAND.

Second:

A third scaling vector, in principle at least, would be to increase the number of bits per cell.

I'm not sure about that. NAND can scale the number of bits per cell, because it's analog (i.e. charge storage). Can SCM be treated as analog? If not, then you only get one bit per cell.

Also, packing more bits per cell would slow down SCM, if it were possible. So, I wouldn't even want it. For me, SCM's main selling point is its speed.

 

[jimmysmitty]

Company: "3DXpoint and other 3D NAND is not the way to go...."

Also Company: "BTW here is our stuff...."

I get their argument but it really feels like they are just doing the normal "My idea is better" marketing.

 

[mradr]

Other than layering - flash NAND has already hit its limit for how many bits per cell we can deal with without it failing over night or getting any slow. Same with getting any smaller as the cell just keeps losing more and more life. If Xpoint can still scale from where it is now - even taking a bit of a hit in speed -it would still out last what flash NAND can do now.

I don't follow why Xpoint can't do more than one layer though - that seems like a false statement? Sounds more like they just haven't explore the idea yet than a real limitation. Heck they could even do layering over layering for both tech and stacking on top of that by building layer one half way - work on layer 2 half way - then finally layer 3 half of layer 2 etc etc till a "stack" and then stack the stacks layer down the line.