What storage form factor due you think will decrease in usage the most over the next 5 years?

[mopardude87]

I don't imagine that sprinkled dog turds would retain data very well during system shutdown. They also would fly apart even spinning @ 5400 rpm. Also I don't see m.2 turds working very well.

I had a friend that once theorized that his cat might make a decent harddrive if you could stand the noise (Kitty Mnemonic).

I heard dogs make very good hard drives but you get the ones that sound like a 74gb WD raptor being flung about the room tied to a shoe string being man handled by a a cow boy who looks like charlie sheen IF he was still on 2 and a half men.

 

[DrMrLordX]

I heard dogs make very good hard drives but you get the ones that sound like a 74gb WD raptor being flung about the room tied to a shoe string being man handled by a a cow boy who looks like charlie sheen IF he was still on 2 and a half men.

So basically like my old 4 GB Western Digital UltraSCSI hdd. Got it.

 

[cbn]

For Intel's ruler form factor it would be interesting to see how the 32TB QLC model (Intel D5-P4326 ) has its 64L QLC packages arranged on the long PCB. With each 1024Gb QLC die yielding 128GB it takes 256 of these dies for 32TB.

So assuming each package has 8 dies that means 32 packages. If each package has 16 dies that means there is 16 total packages.

With this noted, how much extra room on the PCB for extra dies and packages vs. bigger dies/bigger packages vs. how much room in the controller for handling extra dies?

Is the Intel design PCIe 3.0 x 4 controller 16 channel with eight chip enablers per channel with each chip enabler handling two dies. Or something different?

P.S. According to the following picture Intel SSD D5-4326 is capable of having 32TB in both Ruler and 2.5" 15mm form factor:

And looking at the pictures below it appears the Ruler PCB has ~79% greater PCB area compared to 2.5":

So based on that Long Ruler SSDs have quite a bit more room for capacity increase if larger capacity 64L QLC dies were available and/or a controller with more chip enablers were available.

 

[cbn]

Assuming Intel 64L 1024Gb QLC die is ~15mm x ~11 mm (ie, rectagular die like the 1st Gen 32L 256Gb MLC/384Gb TLC die) moving to a ~18mm x 18mm square die (a ~doubling of die area) results in barely an increase in partial dies (24 vs. 22) according to this die per wafer calculator --> https://caly-technologies.com/die-yield-calculator/

Good dies, however, (according to the same calculator) do drop from 294 to 120 though! (However, with Ruler form factor SSD being such a high purpose device I wonder if Intel is still considering moving to such a large die at some point in the future?)

 

[PingSpike]

On the surface it certainly seems like the 2.5" form factor is the least useful. In my mind you have big capacity for cheap. 3.5" wins here, consumer 2.5" hard drives have smaller capacities and worse capacity per dollar than 3.5" offerings. And then you have high speed and "enough" capacity which you'd think M.2 drives should win in.

That said, 2.5" SSDs have better compability as upgrades and their performance disadvantages are unnoticeable in most common workloads. And M.2 drives for some reason come at a price premium that never seems worth it to me. And those 2.5" externals that have $/TB versus 3.5"? They're really popular around me. Turns out, 2TB is a pretty useful amount of space for a lot of users. And they don't require a power brick like the 3.5".

 

[cbn]

According to Flash memory summit 2018 we were supposed to see specs for NVDIMM-P released in 2018:

https://www.flashmemorysummit.com/English/Collaterals/Proceedings/2018/20180806_PreConC_Kulkarni.pdf

I do hope the specs do come soon. (Very interested to see if this tech will allow mixing/combining of Non-volatile memory types on the same DIMM.)