Intel and Micron have announced a new type of memory technology called 3D XPoint that they say could offer speeds up to 1,000 times faster than NAND flash storage.

3D XPoint technology samples will be available later this year, which means we could start to see products that use 3D XPoint as soon as 2016.

3d xpoint_02

According to Intel and Micron, this is the first entirely new type of memory to be introduced since NAND flash was unveiled in 1989.

While 3D XPoint is far faster than NAND flash storage, it’s not quite as fast as today’s DRAM memory chips. But it doesn’t use transistors the way DRAM does, which makes it possible to pack more storage into the same amount of space: 3D XPoint is about 10 times as dense as DRAM.

The memory is also non-volatile: it can continue to store data even when the power is turned off.

All told, Intel says 3D Xpoint offers a fast, cheap non-volatile alternative to existing SSDs, which could increase the speed and capacity of solid state storage while eventually bringing down the cost.

Intel and Micron says 3D Xpoint also has another thing going for it: while NAND storage has a limited number of write cycles, 3D Xpoint does not. That means it could offer up to 1,000 times the endurance as well as 1,000 times the speed. At least that’s the theory. We’ll probably have to wait until next year to learn more about real-world performance (and the price).

Oh yeah, one more thing. Wondering how to pronounce 3D Xpoint? It’s Three-dee Cross-point.

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign


Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Subscribe to Liliputing via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 9,501 other subscribers

32 replies on “3D XPoint storage: Up to 1,000 times faster than flash”

  1. Well, it’s not that easy, to make use of Xpoint the Operating System needs to adapt to the new architecture which I presume is very difficult for Microsoft & especially Apple.. Both the companies are too scared to change & having said that, this could be the chance for Ubuntu, they could become the OS leader.. I mean wow! I’d love to see that happen, I mean, it’ll be great to see money sucking ticks lose for a change!!!

  2. You think Xpoint is something? Take a look at Nano-RAM by Nantero. Fast, non-volitile, 3D high density, low power, durable, and can use existing CMOS Fabs with only minor adaptation. The technology is supposedly proven and now Nantero is licensing the IP to Fabs for testing. Wikipedia has a page on Nano-RAM with links. The technology uses carbon nano-tubes. The concern I have is eventually how do you safely dispose of zillions of old nano-RAM SSD’s? Releasing tons and tons of carbon nano-tube waste into the environment sounds like a very bad idea to me. But heck, I’m no expert.

  3. “1,000 times the endurance as well as 1,000 times the speed”, I want this to market 1,000 times faster than expected!

  4. This is not a pie-in-the-sky tech. This is something that could really destroy the SSD market if they don’t license it.

    1. I don’t think the word vapor means what you think it means. It’s already in the process of being sampled at the end of the year, and there are already wafer outs with the product

    2. Yeah, Intel and Micron aren’t even selling vertical nand flash yet (Samsung has been doing it for over a year, and has 5 year warranties). Intel isn’t even selling 14nm desktop chips yet. Pure vapor.

      1. This is not related to either technology, Plus this is more of a Micron tech where Intel is helping and getting most of the attention.

        1. I’m just questioning the credibility of both companies. I really like my Samsung evo ssd. 256GB for $100, with great performance and increased write cycles.

          1. what does this have to do with SSDs? It’s like saying I like AMD so everything Intel says is not credible.

      2. 1. Neither Intel nor Micron made any mention of what the process node is for this product
        2. Vertical NAND is not the same as this, and if you listened to the press conference you would know. (It won’t be ground breaking if it was vertical NAND anyway)

        1. Intel and Micron announced 3d nand in March, which is identical to Samsung’s vnand. They most likely will have a product on sale before the end of the year… this is not definite, but believable as Samsung proved that those structures are manufacturable and economical. I’m just providing data points of claims Intel has given but not yet made good on them. Only Intel and Micron knows the real details, the rest (including this) is conjecture. At least you learned that vertical nand and 3d nand are effectively the same thing.

          1. Vertical NAND and 3D NAND are of course the same thing, which is to stack planar NAND.

            So do you mean you also said Samsung’s V-NAND was vaper before they came out with commercial products? It’s obvious you don’t know how to differentiate between vapor and not.

          2. Strawman argument. I never said anything about Samsung’s announcements. Which makes your conclusion invalid. Nice try. I commented on track record which might be the most accurate method of vapor detection.

          3. sure. lets go with track record. Name 1 technology (not product, as product lineups change easily but technology roadmap doesn’t ) that Intel or Micron said in a press conference that they planned to implement within a year and did not do it at the end (which means not delayed, but never showed up)
            And my conclusion was valid. YOU SAID “but believable as Samsung proved that those structures are manufacturable and economical” , which means you won’t believe any announcements UNTIL someone PROVED it manufacturable and economical, which means Samsung, the first one to announce they were coming out with it, not believable to you until they proved it.

          4. I gave two originally. 3d nand and 14nm desktop cpus. They are presently not for sale. Intel told investors about 3d nand last year and and their preliminary cpu roadmaps for 14nm had Broadwell desktop cpus predicted for 2014.

          5. delay != vaporware. By your definition, TSMC’s 16nm and 14nm is vaporware too. So is Samsung’s 14nm node, since that’s delayed as well.
            It’s apparent you don’t know that process nodes and fabbing a wafer is really hard work and it’s easy to hit snags that will delay it.

          6. strawman again. I did not say anything about TSMC, or about the difficulty in process development. You asked for an example and I gave 2. You did not comment on the examples I provided.

          7. You know what, you’re just a tool (edited to be more polite)
            Fine, lets go to your 2 “examples”

            “Intel told investors about 3d nand last year and and their preliminary cpu roadmaps for 14nm had Broadwell desktop cpus predicted for 2014.”

            Now tell me where they PROMISED a specific timeframe that a product will come out. Please note the emphasis PROMISED, as roadmaps can change easily. Can’t find proof where they promised? I would be they said “expected”. Now if you want to call that vaporware, then I’m willing to bet there’s tons of stuff that people can tell to you that tons of stuff you said you would do but merely delayed it as vaporware as well.
            Please don’t pretend you know lots about semiconductor businesses, because clearly you don’t.

          8. Google “anadtech Intel 3d nand to ship” and look for the November article. Check the wikipedia article for “broadwell microarchitecture” and look at the section called “roadmap and history”. Those were the sources for which my comments were based.

          9. please, please, tell me that you DO know that a roadmap is not a promise, but a guidance on what can be expected if everything goes as planned. You DO know that, don’t you?

          10. Intel CEO was quoted at 2013 Intel IDF as saying that chips would ship before the end of 2013 (not a roadmap image, but a direct quote). Intel CEO said in a call to investors that 3d nand would ship in 2nd half 2015. These were quotes, not roadmaps. I agree that roadmaps are just dreams.

          11. Sure… Bumpless Build-Up Layers. Originally this substrate technology was announced in 2001 with anticipated use in 2006 or 2007. Then in 2005, it was ‘delayed indefinitely’.

            Here we are in 2015… still no BBUL. Although, here’s a guy who was working on BBUL in 2010 (he’s not the only one, there were tons of job offerings in Chandler for BBUL engineers and technicians in the past 5 years)



            Here’s another one… The Intel LCOS “Liquid Crystal on Silicon” projection TV system. Pretty ground-breaking technology, but was beat out by ever-increasing size of direct imaging LCD and plasma screens. Cancelled in 2004, and the anticipated shipping date for units was 2005. Which was a real shame, as it had a very nice image quality on the demo units I saw.

            I could keep going, but it’s late. 😉

      3. Pure vapor? Not so at all. Samples already in the pipeline.

        I date a CPU designer, and Intel has already been probing the channel for future licensees for this tech.

        This is real and coming very soon.

    3. Apparently you can.

      You have zero idea what you are talking about, but hey, you can repeat a word you heard. Good for you!

  5. Bringing us closer to the day that we won’t need RAM anymore.

    1. Its still slower than RAM, so it will not replace it
      Just like SSDs won’t displace RAM, this won’t either. If it gets close to price parity to SSDs though, it will displace SSDs

      1. What I was referring to is the idea that with faster and faster storage technology, theoretically, computers could operate only on super fast storage, and not have RAM to temporarily handle files.

        The benefit is lower cost, faster performance, and a much lower chance of file-corruption (corruption that normally occurs during the exchange to and from memory).

        I don’t know if this technology is fast enough, but IBM is working on this project. They have a custom OS built for it.

        1. This would definitely turn filesystem design on its head. Not saying that current tech could not work but it’s essentially removing the need for buffers and IO overheads. Not having to move data to RAM to verify data integrity. Things of that nature.

        2. well yeah, if storage can get up to the speed of RAM, then it’ll replace it. But as long as any new type of technology is not faster than some other technology, it won’t replace it.

Comments are closed.