Toshiba-SanDisk claim title to highest-density 16Gb NAND flash

Take that Samsung! Toshiba and their partners in NAND, SanDisk, just announced what they are calling the highest available density of single-chip MLC NAND flash memory. The new 8Gb (1GB) NAND will be available in commercial shipments starting later in Q1 with 16Gb (2GB) NAND on the way in Q2. Presumably, this is the technology at the guts of SanDisk's new 32GB+ SSDs hitting OEMs this spring. Great, but Sammy maintains bragging rights to the fact that they were smart enough to bring 50-nanometer technology to market whereas Tosh tried and failed with a relatively less complex 52-nanometer technology -- Tosh's new single-chip NAND relies on 56-nanometer technology instead. Now, we're not going to pretend to be experts in NAND fabrication, but we fail to see how 56-nm technology produces a higher density than Sammy's 50-nm processes. So who's right? Who cares! Just as long as the trend towards cheaper, faster, and smaller flash memory continues.

[Via AVING]

[zencarver @ Jan 24th 2007 9:29AM]

"Now, we're not going to pretend to be experts in NAND fabrication, but we fail to see how 56-nm technology produces a higher density than Sammy's 50-nm processes."

A larger (56nm) production method can be offset by packaging - different methods of chip stacking, interconnects, etc. Obviously, the 50nm tech can still have a higher theoretical density, but only if you can package it correctly.

[wish @ Jan 24th 2007 9:38AM]

OK, so call me naive, but what's with the nomenclature of Gb vs. GB? Those parentheses threw me off there guys...

And no, the 2G vs. 1GB discussion (how long did that go on?) with the second-gen iPod shuffle did not throw me.

[Matt B @ Jan 24th 2007 9:45AM]

1 bit = 8 bytes
8Gb (notice little b)/8 = 1 GB (Large B, most common usage measure).
Amount of bytes in 1MB = 1024

Ever buy a 80 GB HD and get it home only to see ~72 GB available? It's because of this offset.

[Josh Clark @ Jan 24th 2007 10:06AM]

I think you mean 1 byte = 8 bits.

Also, the hard drive difference has nothing to do with the difference between a bit and a byte, but with the difference between a 2^10 gigabyte and a 10^3 gigabyte (often noted as GiB to prevent confusion, but not by hard drive manufacturers who want to confuse you).

[Rex @ Jan 24th 2007 10:05AM]

I think you mean 1 byte = 8 bits. No worries though.

[kev @ Jan 24th 2007 9:54AM]

There's plenty of logic circuits inside that chip. It's not just space for data...

[strider_mt2k @ Jan 24th 2007 10:02AM]

Oh yeah.
Bring on the lower priced/higher capacity flash-based hard disk replacements!

[Robert Andrews @ Jan 24th 2007 7:17PM]

I think he was saying the offset is because a mega byte is actually 1024 and not 1000.

[Josh Clark @ Jan 24th 2007 3:02PM]

Whoops, now I've screwed up. A GiB is a gibibyte (2^30 bytes or 1024 MiB); GB is a gigabyte (10^9 bytes or 1000 MB). Extremely frequently, however, GB is used when GiB is more correct. Therefore, the hard drive manufacturers are right in giving 80 GB hard drives, it is Windows (and Linux and Mac, IIRC) that is responsible for displaying the size as 72 GB (it really is 72 GiB).

[Matt B @ Jan 24th 2007 10:20AM]

Oops.. Sorry for the mixup. It's early and the coffee has not kicked in yet. :)

[Silver @ Jan 24th 2007 10:50AM]

"16Gb (2GB) NAND on the way in Q2"

Hmm, just in time for the iPhone?

[PaulK @ Jan 24th 2007 12:35PM]

"Now, we're not going to pretend to be experts in NAND fabrication, but we fail to see how 56-nm technology produces a higher density than Sammy's 50-nm processes."

Actually, the bigger issue is that transistors and vias do not behave well at these small sizes. So, the actual logic gate densities are affected more by how the technology gets around the issues and manages errors and all. So, a smaller process size could actually have less density (of gates/logic) due to more complex surrounding techniques. It usually takes 12 to 18 months to improve density with new process.

[flaco148 @ Jan 24th 2007 8:34PM]

Just a guess but the 56nm might be considered denser because it has more memory for its size compared to the samsung 50nm. But samsung might be faster and that is why it isn't as small.

[ConceptVBS @ Jan 24th 2007 9:21PM]

@ flaco148

Dude, you got it completely wrong.

The 50nm would not only be faster but have more memory per square nanometer because you are able to squeeze more transistor into a given space.