Intel Quietly Delays 3DXPoint Memory Modules

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
According to this article, Intel has 'quietly' delayed the introduction of 3DXPoint memory. Most of you will remember that, when we reported on Intel's 3DXPoint technology last year, the company was expecting to ship in 2016 built on a 2-layer, 20nm process in a 128Gbit package. Now that Intel is saying that 3DXPoint memory modules require Cannonlake-EP to work, a late 2018 / early 2019 date is more realistic.

On the Purley platform we're actually starting to sample those products already to some of the leading-edge customers. And they're seeing not only just an overall TCO performance advantage that we typically see with each one of these but this also continues our integration of things like the omni-path fabric, it has more integration of the silicon photonics so it's still the adjacency functions that are quite strong, and they will get more and better as we go through each one of these. There will be a second generation of Purley that includes 3DXPoint. If Intel's upcoming 3DXPoint memory modules require Cannonlake-EP to work, then investors should realistically expect that Intel won't be selling those modules until either late 2018 or in the first half of 2019. I'd be pleasantly surprised if it were the former, as I'm expecting the latter.
 
Intel has a habit of swinging for the fences when making a tech announcement, then using the subsequent months to quietly scale back expectations.
 
Too bad. This seemed like a good product and now it seems like hype. Intel made out to be a replacement to all memory.
 
Intel has a habit of swinging for the fences when making a tech announcement, then using the subsequent months to quietly scale back expectations.
Intel never announced a date for release, and I wouldn't exactly call this announcement of 3D XPoint modules "swinging for the fences":

Intel SSDs will be sold under the name Intel® OptaneTM technology, and will be available in 2016. The upcoming 3D XPoint technology based DIMMs will be available in the next all-new generation of the Intel data center platform.

It's not clear if this is a misunderstanding or an actual delay. The Purley Skylake platform was already on the data center roadmap when the announcement was made, but Skylake and Purley were not mentioned as being the target of this that announcement. It's reasonable to understand Krzanich's quote, "There will be a second generation of Purley that includes 3DXPoint," to be the same as "the next all-new generation of the Intel data center platform". Supporting this is that no one mentioned this as a delay in the transcript Ashraf references, and afaict, it was never announced by Intel for Skylake data center platforms.

To me, this just reeks of more of the problems poor quality rumor writers masquerading as news cause with misinforming people. Plus Ashraf doesn't care if the information he writes is true or not, since he literally lives on click-bait clicks.
 
Last edited:
I think the SSDs are still on track so not all is lost
SSD is good to go to my knowledge.

Too bad. This seemed like a good product and now it seems like hype. Intel made out to be a replacement to all memory.
I think Hexus is just framing it a bit oddly. DIMMs will be a ways out, which I think is understood....
 
Intel never announced a date for release

I'm more referring to the overall technology. Originally it was announced as a game changer. Since then they've been scaling back expectations.
 
I'm more referring to the overall technology. Originally it was announced as a game changer. Since then they've been scaling back expectations.
How is it not a game changer? Other than NAND also going 3D to blunt some of the density improvements, 3D XPoint still has endurance and speed far above NAND, and all products using it are still in various stages of roll out.

No expectations have been scaled back unless I missed some crappy rumor site's dumb proclamation(s), because I don't read crappy rumor sites. If you're disappointed that it's an initial product and not meeting specs announced as long term improvements that were to happen in future generations, I'm not sure what to say about that.
 
My guess is their product they have planned couldn't compete with samsung. But resistive ram or what ever you want to call it is a huge game changer. Its goes way way beyond what 3D flash can do.

Just sucks, cause I was hoping xpoint compete with samsung and bring prices down. Anyway by the time 2019 rolls around 2TB ssds will be around $500 or less imo.
 
I was hoping xpoint compete with samsung and bring prices down.

Samsung what? 3D NAND?? I thought that XPoint was not at all about competing on price with flash but cheaper than DDR3/4 with performance (lower however) similar to DDR3/4 with and non volatile with a much greater endurance than flash.

I was expecting XPoint to cost a few $1 per GB but still way cheaper than RAM.
 
Last edited:
Right now samsung doesn't have any real competition on their latest m2 ssds. Intel was going to be that competition with their Xpoint. Well I was expecting Xpoint to have low end and high end products. But for now samsung is the best for reliability and speed.

but this means resistive ram be delayed another 3 years which sucks.
 
This is ok. Samsung has a faster cheaper solution ready right now. That's the cool thing about competition. If you're not going to bring your A-Game, another company will.
 
My guess is their product they have planned couldn't compete with samsung. But resistive ram or what ever you want to call it is a huge game changer. Its goes way way beyond what 3D flash can do.

Just sucks, cause I was hoping xpoint compete with samsung and bring prices down. Anyway by the time 2019 rolls around 2TB ssds will be around $500 or less imo.

xpoint was NEVER a capacity play vs nand. It has the potential to be a capacity play vs dram, but there are still complications as is to be expected. For the SSD space, xpoint brings significant performance differentiation vs nand though. We've seen real world demonstrations with 1/10 the latency at the application level (which is pretty insane) and significant performance improvements at the application lvl.

Best way to think of xpoint for the next couple of years is how nand was when it first started hitting the server market: high end workload acceleration hardware. Basically, things like the DC P3520 U.2 drives are now the first line storage giving excellent performance at low prices (basically price parity with any SATA/SAS based enterprise SSD but on NVMe with significant performance upside). xpoint will basically slot above that providing write acceleration and first lvl index acceleration.
 
That's a bad shout for Zen, goddammit, shit...

They, seem to me, all they can do, is release Zen, early, before KB arrives....
 
Back
Top