Intel & Micron 3D XPoint Memory

So is this a step in the direction of cartridge games for PC? That is I buy a game on a small chip in a small cartridge and plug it into some sort of port (with chaining) and access the game directly without installing it and it works about as fast as if on a RAM disk. And of course it will be protected from copying so good-bye Steam? And it can be updated via software patches since no one seems able to write a game that works out of the box.

That would be awesome. You could keep a master copy on the disk that isn't writeable and have a copy that you run and keep updated. I hope optical storage dies.
 
Intel is not the sole developer of this technology. And in the field of storage, which this is, there is a good bit of competition.

If this were a new cpu tech the argument would have more weight.

Yeah, Intel and Micron are actually late to the RRAM party:

http://www.dailytech.com/How+Silico...+Beat+HP+to+the+Market+wRRAM/article33122.htm

But that won't slow them down. Intel and Micron were late to NAND as well, but now they're major players.

Right now HP has stopped direct development of RRAM, and handed-over commercialization to SK Hynix Not a top-tier memory maker.

The there's Crossbar, which also has no shipping products after 5 years. Still "IP-only," so no idea how far along their tech is

I would expect us to have to wait 2-3 more years for something we can buy in stores, so really it's anyone's ballgame at this point.

Yup people, this is just a technology announcement, no tests. Nobody is shipping RRAM anytime soon, but it's getting closer :D
 
Last edited:
Yes and no. There's no need for, say, the concept of block size anymore. For stuff that's optimized at a low level (like databases or operating systems), that's major. Also, we don't even have the tools to do much of anything interesting with NVMe's focus on asynchronous data access just yet.

I asked the Intel engineer that was in the Q&A afterwards what was in store on this front, citing the example of what they did with Threading Building Blocks and MKL on the multicore front, and got the "we're just talking about the product today" response :(

Thanks Chris, that was my thought; not necessarily each program needing to be rewritten, but at the very least the OS will need to be written for this specifically. And even the layers above that, to take full advantage, will likely need some changes to take full advantage as well.

Exciting stuff, hope to see more about this from you guys!
 
Pcie4 at +30GBs should be here in time for this tech. If they can saturate a full 16 lanes that would be impressive. Give me that and a cpu with 8+ gigs of HBM.
I was going to suggest that as a joke: PCIe x16 SSD, but that amount of bandwidth is probably an eventuality for some types of storage, especially in enterprise first.

In consumer spaces, NAND storage saturating PCIe 3.0 x4 based M.2 is more a luxury than necessity once it gets there. I think the same will happen with ReRAM storage in the same consumer segment: max bandwidth is less of a concern than lower cost and higher capacity storage. There is likely to be diminishing returns after a certain bandwidth speed; recent M.2 reviews show bandwidth advantages can be diminished by other factors. To a balanced design goal of capacity vs bandwidth... not to (bandwidth) infinity and beyond.

I agree than PCIe 4.0 should give some comfortable headroom for storage bandwidth. I doubt the goal will be to saturate the bus any time soon though.
 
the joint Intel/Micron facility in Lehigh, Utah is actually spelled Lehi, Utah :) I should know I live here.
 
If it can be read and written it can be copied as well. Game developers do not want old style console games because those can be sold freely in physical form. Plus it costs a lot of money to produce vs a digital download.

It's much more profitable for them to distribute digitally and prevent customers from selling their old games to other players.

It could be copied sure, but without the cartridge it won't run. As for selling them, if they don't want that there would be easy ways to stop it e.g. like Windows 10 hardware hash. As for cost, I'm not saying this will be now, sometime in future maybe the cost will be low if this was done. It's soft of like what was done with CDs but faster and less prone to damage (unless you lose the game or drop it and step on it but at least it can't be scratched like DVD's I hate the way the PS4 does this to switch games you have to switch the scratch damage prone Blu-ray disks).

BTW digital download is NOT an option for everyone I get about 1GB/hr and I'm on a faster system than most around here (because I have unlimited 4g with a grandfathered account).
 
The beginning of this patent application has an accessible description of how cross-point ReRAM works: http://www.google.com/patents/US20120170353

Yeah, Intel and Micron are actually late to the RRAM party:

http://www.dailytech.com/How+Silico...+Beat+HP+to+the+Market+wRRAM/article33122.htm

But that won't slow them down. Intel and Micron were late to NAND as well, but now they're major players.

Right now HP has stopped direct development of RRAM, and handed-over commercialization to SK Hynix Not a top-tier memory maker.

The there's Crossbar, which also has no shipping products after 5 years. Still "IP-only," so no idea how far along their tech is
Intel mentions that it and Micron have been working on this type of memory for a decade. The IMFL joint venture is about 10 years old, but I don't see any particular demonstrations by Micron from that time period. Intel did demonstrate a PRAM ("Alverstone", a resistive phase change memory in the memristor family like XPoint) device back in 2006, so 10 years could be accurate.

Twitter had some posts going back and forth among tech editors, and one of AT's editors (IanCutress) said Intel and Micron had solved the problems involved in mass producing this type of memory. There was also direct Q&A with engineers so at least some more details should be coming soon.
 
Yes surely Intel with no considerable competition to speak of, prices the products fairly out of it's kindness. The stock owners are content and are not asking for more.

Being that I've traded $5-10m of INTC and probably $2m of MU for clients over the past year, yeah, the economics of it were my point- the massive capex needed for facilities like fabs is recovered by high prices of early products. Beyond that, they can be more flexible, but obviously everyone is always asking for more. New tech costs a lot to consumers, but isn't an immediate cash cow for the companies that invest in the tools to build it.
 
I dont know what this is but I like this part:

"from 4.4 zettabytes of digital data created in 2013 to an expected 44 zettabytes by 20204"
 
Guys guys guys, this IS NOT going to replace DRAM in your PC ... not yet at least.
-It doesnt have ~unlimited write endurance like regular DRAM
-It is slower than DRAM

Honestly, even though they say it's between RAM and storage, it's going to be a replacement for NAND essentially. You will first see drives that are mostly NAND but have several gigs of this as cache, then you will see full XPoint storage devices. Of course these will come to the enterprise markets first, and then consumer.

It will use NVMe as the command set, and be accessed directly over PCIe, or perhaps via memory DIMMs (like NVDIMMS) but I only really see that being practical in servers.

We wont be getting rid of the split between RAM and storage for a while yet...
 
I would expect us to have to wait 2-3 more years for something we can buy in stores, so really it's anyone's ballgame at this point.

Yup people, this is just a technology announcement, no tests. Nobody is shipping RRAM anytime soon, but it's getting closer :D


Uhh, no dude, they are shipping it already to select customers, the die size is comparable to NAND(128gbit) where as most other RRAM/STTRAM vendors are shipping TINY TINY amounts of storage (tens of Mbits) in their dies. Products will be available next year from Intel/Micron and then the XPRam itself will be on the market next year as well, so other companies will make products too.
 
My biggest worry is the short term affects on NAND based SSD's. We've hit a perfect storm, in a bad way. NAND technology has been pushed down to 16nm, 3D isn't much safer and with a new technology nipping at their heels, some more cost cutting measures may occur to be competitive.

I fear the reliability and bullet-proof nature of NAND drives may be in jeopardy. After all, look what happened to Seagate in the HDD world.
 
Uhh, no dude, they are shipping it already to select customers
No, what Micron is sampling is its first generation ReRAM product. The second generation, which was announced this week as XPoint, isn't due until 2017. Micron's roadmap includes both products as "New Memory A" and "New Memory B": http://www.reram-forum.com/2015/04/01/microns-resistive-memory-roadmap/

The product Micron is sampling is almost certainly CBRAM (filament based ReRAM) which it was developing with Sony and demonstrated in 2014. XPoint is not filament based according to Intel, and neither Micron nor Intel have suggested XPoint was shipping any time soon.
 
No, what Micron is sampling is its first generation ReRAM product. The second generation, which was announced this week as XPoint, isn't due until 2017. Micron's roadmap includes both products as "New Memory A" and "New Memory B": http://www.reram-forum.com/2015/04/01/microns-resistive-memory-roadmap/

The product Micron is sampling is almost certainly CBRAM (filament based ReRAM) which it was developing with Sony and demonstrated in 2014. XPoint is not filament based according to Intel, and neither Micron nor Intel have suggested XPoint was shipping any time soon.

The Anandtech article says otherwise, "The significance of the announcement isn't just the new memory technology, but that it's actually in production with volume shipments scheduled for next year." and "That said, both are working on their own products with first commercial shipments (of the memory itself) scheduled for next year."

From Intel, "3D XPoint technology will sample later this year with select customers, and Intel and Micron are developing individual products based on the technology."

I am fairly certain THIS is new memory type A,not B.
 
You're moving the goalposts. What you said, and what I quoted:
Uhh, no dude, they are shipping it already to select customers

Which is false for XPoint, going by your latest quotes. What is being sampled right now is the filament-based ReRAM Micron was developing with Sony, which was demonstrated last year and fits in neatly with release in 2015 according to Micron's roadmap from a few months ago in April.

If Intel is claiming a 2016 release for XPoint instead of a 2017 (volume shipments), with sampling later this year, then "New Memory B"'s schedule has changed if that's true. But saying parts will be available in 2016 isn't the same thing as a projection of 2017 for volume shipments.
 
Last edited:
Two years later, and still just teasing with 3D renders? Everyone in this thread swore it would be out in 2016, but here we are still waiting :(
 
Back
Top