Micron Unveils 16nm Flash Technology

[Zaxx]

Looks like 16nm is next at bat once 20nm (in short supply atm) dries up. Ofc this mean more chips per wafer which translates into low costs and cheaper (and larger) SSDs. Seems as those the 128Gb die will pretty much be the standard as 64Gbs are now. This paves the way for much larger ssds without choking down 8 channel controllers as we've seen in the past. Biggest draw back is that you'll probably have to buy a 480/512GB or larger SSD to get top performance...similar to the 34nm(32Gb) to 25nm(64Gb) transition which rendered the 60GB 'boot drives' as dogs. I'm hoping the industry can find a way with FW to combat this trend. Imagine having to buy a 1TB SSD just because you want a fast OS/Boot drive...lmao

Here's some sources:

http://thessdreview.com/daily-news/l...-law-rolls-on/

http://www.techpowerup.com/187274/mi...echnology.html

[Kain665]

I wonder how many P/E cycles their 16nm NAND is rated for... 1k? 1.5k?

[Zaxx]

Good question. Forgot about endurance. Heh...watch the eMLC turn out to be 2k...lol

[Kain665]

Too bad the endurance thread died. I guess too many companies got annoyed at their drives' poor showings being exposed.

[Zaxx]

Well the 20nm transition didn't seem to spark much activity. Questioning the new 16nm nand's longevity may be just what it take to start the endurance thread back up. I have an extra rig (C2D E8400 @ 4.25GHz) that I could dedicate to the destruction of an ssd (but I'd have to slap a pci-e to sata 6Gb/s card in it). But just don't have the extra cash, or 'can't justify' is a better term for a 128/256GB ssd to just watch waste away...it's against my 'techie commandments' to allow harm to come to perfectly good hardware...lol

[tiro_uspsss]

Too bad the endurance thread died. I guess too many companies got annoyed at their drives' poor showings being exposed.

what do you mean it died? it was deleted?

I personally hate the 'unstoppable march' toward smaller & smaller nm's - GIMME LONGEVITY DARN IT!

[Kain665]

If you hadn't noticed, the main couple of guys testing their drives just up and disappeared from the thread at around the same time. 20nm should not have stopped the thread as there were well known issues with endurance of the NAND. Also, we had the first TLC drive release as well.

[Zaxx]

lmao...as soon as I start going off about small drives being too slow and getting even slower with 128Gb nand, Samsung totally changes the game by using the nand in 'slc mode' for caching allowing the 120GB ti hit 410MB/s writes!! The Samsung EVO is gonna change the way people think about TLC nand.

Here's the story at Anand.

[canthearu]

As one of the testers in that thread, I'd like to say:

a) Most drives performed very well. While all the drives eventually died, they all did so after simulating decades of normal desktop use. Even the TLC tested ok.

b) Secondly, 20nm isn't going to change the situation for the worse in any appreciable fashion, just like the move to 25nm didn't really change much

c) I'd do more testing, but I'm not in a financial situation where that would be prudent at the moment.

Now, more generally, the EVO's SLC mode is good for burst performance, but performance under prolonged write loads will still be pretty mediocre. For desktops, this doesn't really matter though.

Secondly, whinging about having to buy a 512gig/1TB SSD just for good performance should be cause for a good slapping. 16nm dies will force down the price of NAND per byte, making these large SSDs cheap enough that it doesn't matter. The larger size will also go a long way to counteracting the reduction in write cycles as well.