Enterprise flash storage is proceeding inexorably down the process-shrinking road. But what happens when the shrinkage stops and flash devices evolve from using 2D to 3D chips? SanDisk thinks it might have the answer. SanDisk makes consumer and enterprise flash. Its CEO and co-founder, Sanjay Mehrotra, gave a glimpse into …
I'll settle for 2D memristors
Isn't the interesting question when *2D* resistive RAM will be in production?
Flash gets worse
As Intel makes transistors in CPUs smaller, they get more power efficient and we get more on the die and better processors.
As flash cells get smaller their ability to hold the charge that makes them a viable hard disk replacement gets worse.
Give the drives better and guaranteed life, if I want more space, I'll buy two or three drives. There's not much point in a drive that loses your data.
Re: Flash gets worse
Maybe you should have stuck with ferite core? What was wrong with that?
If you can't see the benefits of smaller and higher density storage...
Re: Flash gets worse
On the other hand, SLC with small cells is probably better on all fronts than MLC even with cells with twice the area. At present MLC and TLC exist because they can't yet make tiny cells right down at the noise threshold for single-bit reliability.
My main question is.. how are they going to get rid of the heat?
Re: 3D semiconductors
Getting rid of the heat? Are SSDs anywhere close to overheating at present? A 256Gb SSD with a 2.5 inch drive footprint runs on about 0.5W and is scarcely warm to the touch. A smartphone CPU uses more watts in a footprint of a square centimeter.
The first step that they don't yet need to take would be passive heatsinks on the flash chips, like a low-end fanless GPU or top-speed RAM. Active cooling wouldn't be a problem for a datacentre solution - after all, the CPUs are all actively cooled.
Not to say that memristors won't wipe out flash in the longer term!
2-3 years away would not be 2013 as the story sub-title claims
The problem with stacked ICs is heat concentration and insufficient cooling. The failure rate also increases meaning unit costs increase. We'll see how this all shakes out down the road.
Re: 2-3 years away would not be 2013 as the story sub-title claims
It does say 2-3 years away from mass producing. Just because they get some output this year, that is a far cry from mass producing.
These aren't stacked ICs, they are single ICs fabricated to have 3D structure, and I am quite sure they know all about the heat and any other like issue.
One of their production managers..
I spoke with one of their production managers, a Mr. A. Square of Abbot Lane, who told me "The hardest part about all this was convincing the higher-ups that there was indeed a third dimension to begin with - I finally had to bring in a consultant acquaintance of mine, Mr Sphere, to give them a presentation on the subject. After that, there was a bit of back-and-forth, but finally they bought in."