Western Digital BiCS3 Flash Goes QLC - 96GB per die!

Subject: Storage | August 2, 2017 - 06:21 PM |
Tagged: BiCS3, western digital, wdc, WD, tlc, slc, QLC, nand, mlc, flash, 96GB, 768Gb, 3d

A month ago, WD and Toshiba each put out releases related to their BiCS 3D Flash memory. WD announced 96 layers (BiCS4) as their next capacity node, while Toshiba announced them reliably storing four bits per cell (QLC).

View Full Size

WD recently did their own press release related to QLC, partially mirroring Toshiba's announcement, but this one had some additional details on capacity per die, as well as stating their associated technology name used for these shifts. TLC was referred to as "X3", and "X4" is the name for their QLC tech as applied to BiCS. The WD release stated that X4 tech, applied to BiCS3, yields 768Gbit (96GB) per die vs. 512Gbit (64GB) per die for X3 (TLC). Bear in mind that while the release (and the math) states this is a 50% increase, moving from TLC to QLC with the same number of cells does only yields a 33% increase, meaning X4 BiCS3 dies need to have additional cells (and footprint) to add that extra 17%.

The release ends by hinting at X4 being applied to BiCS4 in the future, which is definitely exciting. Merging the two recently announced technologies would yield a theoretical 96-layer BiCS4 die, using X4 QLC technology, yielding 1152 Gbit (144GB) per die. A 16 die stack of which would come to 2,304 GB (1.5x the previously stated 1.5TB figure). The 2304 figure might appear incorrect but consider that we are multiplying two 'odd' capacities together (768 Gbit (1.5x512Gbit for TLC) and 96 layers (1.5x64 for X3).

Press blast appears after the break.


Further Enhances Industry Leadership in Multi-Level Cell Storage Builds on X4 Expertise in 2D NAND

SAN JOSE, Calif. — July 24, 2017 — Western Digital Corp. (NASDAQ: WDC) today announced its successful development of four bits per cell, X4, flash memory architecture offering on 64-layer 3D NAND, BiCS3, technology.  Building on its pioneering innovation of X4 for 2D NAND technology and past success in commercializing it, the company has now developed X4 for 3D NAND by leveraging its deep vertical integration capabilities.  These include silicon wafer processing, device engineering to provide sixteen distinct data levels in every storage node, and system expertise for overall flash management.  BiCS3 X4 technology delivers an industry-leading storage capacity of 768 gigabits on a single chip, a 50 percent increase from the prior 512 gigabit chip that was enabled with the three bits per cell (X3) architecture.  Western Digital will showcase removable products and solid-state drives built with BiCS3 X4 and systems capabilities in August at the Flash Memory Summit in Santa Clara, California.

“The implementation of X4 architecture on BiCS3 is a significant development for Western Digital as it demonstrates our continued leadership in NAND flash technology, and it also enables us to offer an expanded choice of storage solutions for our customers,” said Dr. Siva Sivaram, executive vice president, Memory Technology, Western Digital. 

“The most striking aspect in today’s announcement is the use of innovative techniques in the X4 architecture that allows our BiCS3 X4 to deliver performance attributes comparable to those in BiCS3 X3.  The narrowing of the performance gap between the X4 and X3 architectures is an important and differentiating capability for us, and it should help drive broader market acceptance of X4 technology over the next several years.”

This latest achievement follows a nearly three-decade long legacy of industry firsts in flash innovation, including the industry’s multi-level cell (MLC) flash technologies using two bits (X2) and three bits (X3) per cell.

The company expects to productize its 3D NAND X4 technology across multiple end-use applications that can take advantage of the higher capacity points supported by X4.  Future generations of 3D NAND technology, including the 96-layer BiCS4, are also expected to feature X4 capabilities.

About Western Digital

Western Digital is an industry-leading provider of storage technologies and solutions that enable people to create, leverage, experience and preserve data. The company addresses ever-changing market needs by providing a full portfolio of compelling, high-quality storage solutions with customer-focused innovation, high efficiency, flexibility and speed. Our products are marketed under the HGST, SanDisk and WD brands to OEMs, distributors, resellers, cloud infrastructure providers and consumers. Financial and investor information is available on the company's Investor Relations website at investor.wdc.com.


Western Digital is a registered trademark or trademark of Western Digital Corporation or its affiliates in the U.S. and/or other countries. All other marks are the property of their respective owners. © 2017 Western Digital Corporation or its affiliates. All rights reserved. 

1 – 1 gigabit = 1,000,000,000 bytes. Actual user storage less.

Video News

August 2, 2017 | 07:44 PM - Posted by Mobile_Dom

Interesting. a low end, potentially DRAM-less NVMe SSD with QLC NAND should hopefully get us to the sub 0.10c/GB future we need to get to.

Or we stick with SATA-III, because it's fast enough for most people and is a metric fuck tonne cheaper (that's 3 Imperial fuck tons if my conversions are correct) than NVMe ones are.

I also just think to be able to say that a laptop has a 500gb SSD inside is going to be a boon for lower end laptop sales. Sure it will be a SATA-3 QLC drive, but it'll still be faster than the 5400rpm mechanical drive the OEM would have put in before this.

August 2, 2017 | 08:29 PM - Posted by pdjblum

but the price doesn't really go down, just the quality

i am an mlc snob because i am maybe too concerned about data loss and i just don't mess with TLC, and certainly not QLC

this reminds me a bit of when all you could buy was a 1080p while the 1440 pretty much disappeared

fortunately people realized they were getting suckered, and we can once again enjoy screen resolutions that make sense and then some for people who do any sort of real work on the pc, not to mention gaming

i think apple has always realized the importance of high res high quality screens and now high end nvme

August 3, 2017 | 10:58 AM - Posted by razor512

Agreed on the price and quality, with TLC and QLC, there are many claims of how much additional data can be stored, but it almost never leads to an equivalent reduction in the cost per GB. at best, you will see a 5-10% reduction in price for a 70+% reduction in write endurance.

August 2, 2017 | 08:54 PM - Posted by OldButNotFooled (not verified)

"1 – 1 gigabit = 1,000,000,000 bytes. Actual user storage less."

Vas is dis? It makes no sense, 1 gigabit = 125 megabytes.

I'll take my memory sizes in powers of two, it's a computing product is it not! So just make things in Kilobyte(1,024) multiples.

And that crappy term "Mebibyte" that they came up with in 1998 is total crap in computing as a megabyte should still be called 1,048,576 bytes. Silly storage industry they are the ones who should have been forced to use "Mebibyte" for their 1,000,000 storage bytes.

As for me it's flash cell states of three leave it be and flash states of four I'm out the door! I'd rather have SLC or none at all and stick with spinning rust! Hopefully 3D XPoint and hybrid XPoint/Spinning Rust drives will come onto the market with at least 128GB+ of XPoint cache to put down all this NAND nonsense!

August 2, 2017 | 09:46 PM - Posted by Allyn Malventano

Yes, their press release had that glaring error (in addition to the whole GiB vs. GB vs. Gib vs. Gb thing).

August 3, 2017 | 09:56 AM - Posted by Rexyl (not verified)

That's not how standards in science work. Base 10 prefixes are base 10 prefixes. Storage vendors are technically correct which is the best kind of correct.

August 3, 2017 | 11:58 AM - Posted by Anonymously Anonymous (not verified)

Don't tell that to Australia, their laws are more correct than anything.

August 3, 2017 | 04:50 PM - Posted by OldFlatulence (not verified)

Yes but in computing everything is done in binary base 2 and converted to base 10, and in computing the term K stands for 1024, and not 1000, hence the term Kilobyte is still defined as binary bits converted to decimal(1024 base 10). That "K"(8K/whatever K of memory) term was a computer sciences memory, and a storage size term, before it was appropriated by the human resources(headhunters advertising for computer positions) folks to describe salaries in terms of multiples of 1000(“hiring computer programmers 100K+ salary potential”). And storage vendors are the ones responsible for the 1998 naming nonsense.

In computing, storage sizes where listed in binary converted to base 10 before 1998 and that storage industry obfuscation business. Look at all the Hard-Drives and those sector sizes and the block sizes are all base 2 derived( 512 to 4096 ). And on the IBM mainframes programmers could set/tailor their own formatted block and sector sizes and hard disks could be formatted with whatever size suited that data that the programmer was using for block/sector sizes, including the ability to request cylinder mode allocation for files on milti-platter hard-drives so files are allocated stored top to bottom on across vertical groupings("Cylinders") of platters for seek time efficiency gains. Programmers could request whole slices of hard drive disk packs be allocated/custom formatted to most efficiently manage their needs, and seek time, and other metrics where much slower/more affected by latency back then. Then there was the question of VM paging file swap space under VM 360/370 where everyone(Advanced assembler students) had their own VM 360/370 OS instance provided so they would not bring down the entire college's computer when their assembly language code ABENDed or crashed the VM instance.

Each of the sciences/disciplines has their own terminology and naming nomenclature and that has not changed it’s only because of the PC and computers in the hand of the untrained that that marketing dumbing down/obfuscation of computing sciences terminology/nomenclature began. That and the HR IT/programming headhunter folks in the late 1970s into the 1980s-1990s and beyond started the salary “K”(multiples of 1000) nonsense. Counting in computer sciences starts at 0 and not 1, it's all 0 relative because 0 is a number/address number/register number/processor number/etc. etc!

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

This question is for testing whether you are a human visitor and to prevent automated spam submissions.