New NAND Storage Development: SLC, MLC, TLC, QLC NAND ... going for PLC

Wonder how many layers they can get this down to before reliability and the hardware's overall lifetime is reduced too badly, Penta(gram) layer SSD now and next up I suppose is hexa, whatever seven is called and then octa if this keeps going on.
Then again really long huge numbers shrinking doesn't really say much either, hardware dies whenever the user is most inconvenienced by it as per some law from how the universe just works and that's that.

Suppose storage capacity and affordability can continue to go up too, just have to monitor overall status and keep the priority files backed up and that should be good enough.

EDIT: As for regular usage I'm pretty sure 3 - 5 years is still more than feasible same as the general lifetime for a HDD before they start showing signs and even then it can last quite a while plus no mechanical parts in a SSD and it can theoretically keep going even more removing that little wear and tear and other issues with spinning headers and platters even if the memory cells have a finite lifetime but tech keeps it optimum so it gets a lot of lifetime regardless of some arbitrary number of hours.

 

If it means I can get a 4TB games SSD for ~$150 I’m all for it.

 

Development of new technologies is always welcomed.
That said, I hate the reliability of QLC and much rather stick to TLC. I can't imagine the endurance PLC will have...
I'm sure that with help from development of PLC, QLC will get better reliability. But there's a point that I'd rather my drive live for longer than having more density.
My current Crucial MX100 256GB drive has 5 years and still going strong. I'm upgrading to NVMe but this drive will live on my old mom's computer to keep working.

 

if we look at this, basically nand maker somewhat "fooling-around"
at beginning SLC - high-durability / expensive, so to push price lower, they make QLC/PLC - low-durability / much cheaper

luckily the nand capacity become so big, basically it cover-up the cons that qlc/plc have which is low write amplification, and as most user wont full-up their storage having free-space also help

SLC = 100K P/E cycles
MLC = 10K P/E cycles
TLC = 3K P/E cycles which i still think sweet-spot
QLC = 1K P/E cycles
PLC = 300 P/E cycles?

even SSD-maker using SLC cache system, far-better controller etc.
if we keep going on this trend, somewhat it make me think, probably the future SSD durability wont be much different than traditional HDD, especially in write-intensive environment

 

Usually you have most drives starting to have issues with non-nand related stuff (controller etc), and around 1 out of 4 drive (of the same type)
will have nand related problems after a few years, so it might work for most consumers (especially at lower price),
but when i saw that the sustained TLC numbers on a nvme get beaten by an older MLC drive running sata,
yeah, i wont get anything "after" TLC.
even switched back to MLC for one of my backup drives i replaced, just to make sure.

 

What I don't really get is why stop there? Why not just jump to OLC (octa level cell)? That way you get a nice even 1 byte per cell. I'm not sure if that affects performance at all but I imagine it'd be easier than an awkward number like 5.

Same. Most of the detriments to such SSDs are totally irrelevant to things like game and media storage.


At some point, I'm going to need to upgrade my RAID1 array in my home server, which uses a pair of laptop HDDs. Ditching them for a low-cost PLC SSD is really starting to sound enticing. I'm basically just waiting for one of the drives to die, or, until I run out of space, so I have an excuse to ditch them.

 

I just finished watching this video and i was shocked when i've seen copying files onto QLC drive is comparable to/worse than copying similar files of similar total size between HDDs. On HDD similar process goes at speeds between 100-150MB/s (depending on particular models or units) and speeds are stable all the time, unlike as with this QLC drive.

 

Getting closer & closer to yearly disposable drives with the big jumps with QLC & now PLC.

 

Do you Yakk your SSD a lot?

 

QLC only slows down if you're copying really big files, and the SLC cache gets used up. Reading is still a lot faster than mechanicals.
I switched to a quiet case not long back and the mechanical HD was the last noisiest bit, so swapped that up for a QLC SSD once it started making strange noises.
You still get better slower degradation rather than complete fail with mechanical drives, and the quietness of them is appealing to me at least.

 

I didn’t click the like but I’m assuming it’s the Linus video. Keep in mind he used an unrealistic scenario to slow the drive down. Most people won’t be simultaneously copying 10GB files from one drive to another on a daily basis. As I and others have stated (even Linus in the video) this a a perfect game drive.

 

It's from Christopher Barnatt (ExplainingComputers). I really like his channel. Linus is fine, but can be a drama queen sometimes.

 

@Richard Nutman
so far i had never any problems with degradation, but the controller dying, and no data recovery possible.
the only HDDs i couldn't recover, were the grinding/clicking units, were customers tried to boot/mess with it, even that it made a strange sound..

 

That's unlucky with the controller dying. What brand(s) was that ?

 

one was a myssd msata (only port that was left on a matx board) and the other a ocz vertex4.
both were mainly OS drives (but still some user data i wanted), so not too bad,
but i wont care to reinstall all my games (siege alone is around 120gb already), just because i wanted a big cheap drive.

for me perf does matter, and after tlc, the sustained speeds are below what i get from a 7200rpm 2.5 hdd, even my 960 evo will have lower writes than a mlc sata drive i use as external backup,
so that will be the type i wont go past..

 

if u look at google there tons report of dead(complete-fail) SSD too, i personally read more dead ssd rather than sudden dead hdd without any sign
eitherway complete fail in hdd is same to various chip failure in SSD which fail completely the SSD

if u read most hdd reliability report, complete hdd failure is quite low, usually hdd will giving sign first, like sector error, read/write error etc. before complete failure.
thats why like backblaze, not straight replace the hdd, when something flagged in smart status

so overall for me personally, both SSD & HDD is not perfect, both have cons and pros
if only SSD keep on SLC/MLC with capacity+price like now, then it would be ideal storage system

even that ssd still not good for archival purpose, not that hdd perfect either, there is data-degradation/decay in all storage medium, but compared ssd, hdd still hold better

 

Thing is that for some weird reason, it is not so hard to find TLC SSDs at almost same price as QLC. I can't really justify QLC.

And funny part is something like: Kingston SSD A2000 1000GB
It is TLC, is quite cheap (SATA TLC cheap) and fast enough to saturate 2nd M.2 slot on x470 board. Would that be 2TB at appropriately higher price, I would already have that.

 

This is my personal experience - mechanical HDD's tend to die slowly and you normally have time to do something about it before they fail. An SSD is going fine one day, and dead the next, which is very annoying.

 

I got the Samsung 1TB 860 QVO. It's the 2nd cheapest 1TB SSD on overclockers right now. Performance is still way higher than mechanical in real world use, read speed still saturates SATA.

I expect by next year you should be able to get 2TB QLC for the same price.