back to article Intel XPoint over-selling criticism surges as Chipzilla hits back

You might have noticed that El Reg has suggested Intel and Micron’s non-volatile XPoint memory claims are a tad overblown – for example here and here. Now Charlie Demerjian is giving Intel a good kicking in SemiAccurate, alleging the original claims were inaccurate and that “Intel’s Xpoint is pretty much broken”. For example, …

  1. Novex

    Intel and Micron's Bluff Is Being Called

    Intel and Micron's vapour announcements are beginning to dissipate, and it looks like they were stretching the truth a 'little'.

    Ultimately we'll know when product hits the streets properly, and in particular when XPoint Dimms debut with systems designed to take them. Of course, by then Nantero's carbon nanotube NRAM might manage to make the light of day and wipe the floor with it (again, if their vapour announcements are anything to go by). I do wonder if the Intel/Micron XPoint announcement last year was an attempt to get ahead of what Nantero were doing.

    1. BillG
      Holmes

      Re: Intel and Micron's Bluff Is Being Called

      El Reg notes this sentence in particular: "System level performance is determined by the full collection of components, with media being only one of these components."

      Well, duh.

      Look, if it turns out it that XPoint is 10x faster than NAND Flash instead of 1000x faster, then people on El Reg will still buy it and not care what was first claimed by Intel because 10x faster is still great news.

      From my view on the inside, it seems that at Intel the tail is trying to wag the dog (Marketing is trying to drive engineering instead of the other way around). Intel used to have brilliant marketing - now, Intel marketing believes whatever it wants to believe, including ignoring the reality Intel engineering is telling them. So someone in Intel engineering tells Intel marketing, "It should be 10x faster but it might be 1000x faster.". Guess what marketing prints? Then marketing dumps on engineering because "you told me it would be 1000x faster, didn't you?"

      While I was writing on an article the latest Intel Quark, Intel marketing kept insisting that I put a spec in my article that blatantly contradicted the datasheet, a spec that Intel engineering also told me was wrong.

      BTW based on what I have read from Intel, XPoint looks like ReRAM to me. It will be late, but it will work as soon as Intel solves the yield problems.

      1. Anonymous Coward
        Anonymous Coward

        Re: Intel and Micron's Bluff Is Being Called

        "We might say that the original XPoint claims referred to raw media comparisons and not system-level performance"

        But how does 100x endurance at the raw component level translate to 3x endurance at the system level? And what type of flash are they comparing to anyway? This is just BS.

        I suppose if Intel had originally said "yay, we have this brand new non-volatile memory technology in the lab! It'll be out sometime in the future and it will have 3 times better endurance than MLC flash!" then nobody would have taken any notice of it. Rightly.

  2. Anonymous Coward
    Anonymous Coward

    Seeing "up to 10x throughput increase and 4x latency decrease"

    Does "up to" mean those are the peak readings with datasets that are worst case for flash? Or were those the best they did against the various flash SSDs they compared with? Speaking of whose flash are they measuring against, Intel's? The consumer or top tier enterprise stuff?

    Their statement doesn't fill me with a lot of confidence. I guess we will have to wait until someone like StorageReview or Anandtech gets their hands on it to see if it is even any sort of an advance over the best SSDs can do!

  3. Nate Amsden

    kind of funny

    or sad to see some folks here on el reg (in the forums) touting how great Xpoint will be every so often yet expectations keep getting lowered.

    For me I'm really not paying attention until the product starts shipping and people see what the real thing is. Until then what's the point of getting excited ? SSDs and NVMe flash are already so damn fast, and durable.

    My own org has been using a 3PAR 7450 for nearly 24 months now with what Sandisk would call "Read intensive" SSDs (though HP has no limits as to workloads for their 5 year warranties). After nearly 24 months of usage across hundreds of VMs and MySQL servers the oldest SSDs in the array have 98% of their life left in them according to their internal metrics anyway.

    There may be a few extreme edge cases where something like Xpoint will really make a big difference but customers already have a lot of choices for very fast storage. I didn't get the impression that Xpoint will be dramatically cheaper either, which really would be more of something to get excited about I think.

    I have a pair of Samsung 950 Pros (NVMe) in my Lenovo P50 laptop(which the Samsung windows tool says has PCIe 4x with 10Gig interface w/ i7-6820HQ CPU), I see absolutely ZERO difference in real world performance between those and the Samsung 850 Pro (2.5" SATA) SSD. Feels kind of crazy to have 3 SSDs in one laptop but I figured what the hell it has the space so I put them in.

    I would wager if I replaced all of my SAS SSDs in my 3PAR arrays with NVMe I would see no difference either, because well the individual apps just don't drive that much I/O (much of it is handled in memory caching). Sub millisecond response time is more than good enough for any of the VMs or databases I have worked with over the past 15 years.

  4. Anonymous Coward
    Anonymous Coward

    I contacted my sales guy at Intel ages ago to ask about getting on the X-Point bandwagon, the prospect of a large volume of chip sales didn't get their attention and I was disappointed. Perhaps they've been distracted?

  5. Anonymous Coward
    Anonymous Coward

    Data retention numbers?

    I'd also take a close look at data retention numbers on the media. Little birds are tweeting that it is only a few days for the supposedly non-volatile memory..

  6. P0l0nium

    Lots of opinions but .....

    ... No data !

    Now who'd a thought it? ... on an El Reg vox-pop !! :-)

  7. Joerg

    The real question here is ... who is behind all these rumors telling that the Intel XPoint would be a huge disappointment and Intel just kept lying , uh?

    Competitors indeed.

    Competitors selling the unreliable NAND Flash crap SSD do not want Intel/Micron 3D XPoint technology to succeed because if that happens they would start losing money and would have to quickly spend billions trying to come up with a competing technology as fast and reliable as the 3D XPoint is going to be.

    The SSD NAND are overpriced unreliable mostly defective products. The industry is full of cheaters and thieves selling the worst.

    1. isogen74

      SSDs didn't really change anything, other than adding a few interesting catastrophic failure modes when their firmware goes AWOL. Mass storage drives are always unreliable, and have been since they were first invented. Stuff on them is nearly always valuable. Therefore you have a decent mirroring / backup strategy, with recovery time dependent on the value / importance of the data.

      XPoint won't change that - data loss is almost never an option, and even if the drive doesn't fail, the building might get hit by lightning, catch fire, etc ... so you still need backups and second drive.

      Honestly there isn't really a distinction between a good drive and a bad drive - the difference between 99.99% and 99.999% reliability is immaterial if your business *relies* on that data you can't afford to be the people in the bad luck zone (either because of bad drive, act of God, or human error).

      > as fast and reliable as the 3D XPoint is going to be.

      Reliable. Irrelevent - see above.

      Fast - either provide data, or sit quietly and wait. We've been promised the "next big thing" in both DDR and HDD replacements for at least the last 20 years. Most of it is expensive vaporware which never gets out of R&D and into mass production; even if the numbers are great, if you can't mass produce it on cost and with high yield then it's DoA.

    2. JohnyDoe

      So true

      I cant believe they are 5 monkeys putting negative vote on your claim. I also cant believe people are whining now. Are you all clowns insane? You still get 4x and 10x. Who has that? They were measurements made on Optane few months ago at the University of New Hampshire’s InterOperability lab. Quote from the article. "However, SSDs based on 3D Xpoint will be too fast for traditional SATA or SAS SSD connections, which would easily limit the drive’s performance. It is being speculated that Intel helped to design the NVMe interface because the company is working on a new memory type." Plus, Intel is facing limitations from some of the controllers.

  8. luis river

    very far

    Prediction on time frame 3D xpoint (massive production) and my offer is at least 24 months (very optimistic) I do not believe that wrong

  9. Anonymous Coward
    Anonymous Coward

    The Long Term View

    My understanding (and I speak as someone who has signed an NDA on this, hence AC) is that the enormous figures refer to what will eventually be available rather than what the first generation of products will be able to do.

    1. Alan Brown Silver badge

      Re: The Long Term View

      My understanding (as a purchaser) is that Intel made these claims about what the first release would be capable of, sans wrapping in logic to make it a drive.

      In any case, as a purchaser there are 2 options: "Buy" or "Not buy" - and that will be dictated by the actual real world performance and price of the release product - which is going to have to be _very_ good and on par with equivalent flash respectively - bearing in mind that once there's another technology in the market flashmakers are likely to respond by lowering prices a bit.

      Being twice the speed for 10 times the price won't cut it - and the kinds of buys who want that kind of performance are going to be _extremely_ sensitive to latency.

      If Intel tells porkies about performance of the released products then that will catch up with them _very_ quickly and Xpoint will become yet another casualty on the road of solid state storage. History shows that companies which make wildly optimistic statements about their pending products don't do very well out of it. This kind of thing could end up driving Intel out of the sector entirely.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like