According to an announcement made early this year by Intel and Micron, their 3D XPoint joint venture will soon be no more. For some years now, the two companies have collaborated to co-develop 3D XPoint memory technology. However, sometimes long-term partnerships get disbanded, and the one between these two tech innovation giants will be dissolved after the release of the second generation of 3D XPoint near the end of 2018 and beginning of 2019.
3D XPoint is a disruptive memory technology that offers thousands of times faster and better storage than DRAM and NAND. And although both companies share the underlying 3D XPoint material and will continue to manufacture the technology at the co-owned Intel Micron Flash Technologies (IMFT) facility in Utah, the plan is to eventually pursue the third generation of 3D XPoint and other future developments independently.
Plainly put, after the breakup, they will uphold unique strategies, designs, and developments for upcoming 3D Xpoint innovations and will market the end products under different brand names. Intel’s products will be branded as Optane, while Micron’s will be known as QuantX.
It is crystal clear that both companies want to offer storage and data centre memory solutions—but in different directions. For instance, Intel wants to continue focusing on using 3D XPoint as a replacement for DRAM, while Micron is anticipated to develop 3D XPoint memory for use with future SSDs and other storage device generations. By going their separate ways, both will be able to optimize the emerging technology and align it with their product roadmaps and business needs while benefiting customers and shareholders alike.
A Disruptive Nonvolatile Memory Technology
3D XPoint is a disruptive nonvolatile memory technology designed to provide lower latency and even greater endurance than NAND memory. It features super-fast and inexpensive memory to provide high-performance levels and deliver low-latency reads and writes.
Allow me to give you a more vivid picture of what 3D XPoint is all about. Imagine any device that requires computing memory and storage needs out there and its current response time. Well, now imagine that response time 1000 times faster. 3D XPoint offers such speeds, and Micron even claims device latency with 3D XPoint will scale in nanoseconds. On top of that, 3D Xpoint is a purely non-volatile memory—and that’s what makes it revolutionary.
3D XPoint Technology Architecture and Performance
3D XPoint technology features an innovatively simple, yet stackable, transistor-less design architecture that creates a three-dimensional checkerboard. Here, memory cells sit between bit and word lines, allowing individual cell interaction. As a result, data can be read and written in small sizes, enabling blazing-fast read and write processes. Bit storage is established on a stackable c ross-gridded data access array.
When it comes to performance, 3D XPoint has the capability to enable high-speed and high-capacity data storage, promises to enable entirely new applications, and creates new possibilities for system architects.
With up to 1000 times greater endurance than NAND, and lower latency, 3D XPoint delivers high-level performance for big data applications. HDD latency is measured in tens of milliseconds, NAND latency is measured in microseconds, and 3D XPoint technology latency is measured in nanoseconds.
By the time an HDD sprints a basketball court, NAND has finished a marathon, but 3D XPoint technology has virtually circled the globe. Currently, Intel employs 3D XPoint to develop cutting-edge solid-state drives (SSD), with plans to apply the memory as a reasonably priced DRAM replacement for enterprise applications.
3D XPoint Technology Memory Innovations
Several memory innovations make 3D XPoint inexpensive, fast and non-volatile, including the following.
Cross Point Array Structure
The cross point array compact structure is the reason behind 3D XPoint’s density and high performance. Bolt-upright conductors link to approximately 128 billion tightly packed memory cells that store a single bit of data.
Current technology stacks 128 GB across two stacked memory layers for every die. If I jump into the future generation’s 3D Xpoint, there is amazing potential to increase the memory layers or even employ conventional lithographic technology to upsurge die capacity.
By changing the amount of electrical current directed to every selector, memory cells can be written or read. It is this procedure that increases memory capacity, reduces costs, and does away with the need for transistors.
Super-Fast Switching Cell
Because of the tiny memory cell size, a super-fast switching selector, cross point compact array structure, and a really fast write procedure, the memory cell can shift its states much faster than any other already existing non-volatile memory technology.
Before 3D XPoint Technology
For decades, DRAM and NAND have been the prevailing middle-of-the-road memory technologies. DRAM is the memory accessed by the CPU quickly, and NAND is for solid-state storage. Although their cell designs have progressed through the years, enabling up to 20nm and lower scaling, the basic dynamics behind the way DRAM and NAND work haven’t experienced any notable change, which makes the two technically limited.
Take DRAM, for instance. It truly lowers latency and can boast unlimited endurance, but then again, this comes at a huge cost of high power consumption, enormous cell size, and cell volatility. What’s more, DRAM requires occasional cell refreshing. That said, cells are unable to retain data in an off state, and that makes the memory unfitting for permanent data storage.
Conversely, NAND has a more efficient cell structure, provides higher latency with the limited number of write cycles, is non-volatile, and enables low-cost memory suitability. Integrated at the system level, both DRAM and NAND make a great team, which explains why modern computers use DRAM as a cache memory and NAND for storage.
But still, latency and capacity gap problems exist. This is where 3D XPoint comes in to save the day. For applications still relying on NAND, 3D XPoint technology already eradicates potential endurance challenges. But again, DRAM scores high points because of its infinite endurance capability. My prediction is that 3D XPoint will potentially replace current mainstream memory technologies if it can provide more endurance than DRAM. Intel and Micron may have placed it on our tables, but for now, we have yet to bid DRAM goodbye.
Are you thinking that the future of fast storage and memory is aeons away? In my view, Intel and Micron have already presented it to us. If I judge by the record of technology innovations, expertise, and development that the two have, they will continue to innovatively drive the next future generations of memory technologies. I even think that their partnership dissolution will make the future quickly materialize—because there will be competition.
This may sound smooth, but in the end, I see implications that need to be dealt with as a result of the dissolution. Both companies will have to dig deep into their pockets to cater to individual research and development costs, internal engineering team’s replacements, resources, and collaborative efforts, which is easier said than done. But for now, it’s interesting to speculate on how this split will affect memory in the long run—hopefully for the better.