Intel 3D XPoint NV memory is up to 1,000 times faster than NAND

[Kumagawa]

So what you're saying is that they shouldn't try and make a new modern, fast, power saving memory? And simply keep on revamping 30 year old tech? Living in the past, eh?
I think this looks awesome. True SSD replacements (1TB and above) could be the size of Compact Flash. No more worrying about only having 3GB of phone RAM, etc etc future yay. Bring it on I say. These Boffins are awesome.

Not at all it's just I've been reading about new memory tech for over a decade none of which has ever left the lab.

[ed^chigliak]

Maybe new hybrid drives with 16-32GB of this stuff combined with 500GB - 2TB NAND. I imagine something like this could be made affordably expensive for enthusiasts.

[shaithis]

Not at all it's just I've been reading about new memory tech for over a decade none of which has ever left the lab.

Like SSD?

[KeyboardDemon]

Maybe new hybrid drives with 16-32GB of this stuff combined with 500GB - 2TB NAND. I imagine something like this could be made affordably expensive for enthusiasts.

That doesn't make sense, I thought that the new 3d Xpoint modules were supposed to offer cheaper, faster and more efficient storage over Nand, if that's the case why would anyone want to combine 3d Xpoint with more expensive, slower and less efficient technologies, it wouldn't make the overall production costs cheaper nor would it improve performance, as far as I can tell, doing this would actually introduce a new bottleneck.

[Scow Captain]

Oh goody another paper/lab only memory format to throw on the pile.

The very first sentence of the article says it's being manufactured now! Intel have said it will be available in 2016 and part of the Xeon Purley platform in 2017. That may well mean 2018 for mainstream consumers but that's three years away not 20-30 years.

[DanceswithUnix]

That doesn't make sense, I thought that the new 3d Xpoint modules were supposed to offer cheaper, faster and more efficient storage over Nand, if that's the case why would anyone want to combine 3d Xpoint with more expensive, slower and less efficient technologies, it wouldn't make the overall production costs cheaper nor would it improve performance, as far as I can tell, doing this would actually introduce a new bottleneck.

Not quite, it is faster and possibly more energy efficient. If it was cheaper, I don't think Intel would be interested

Charlie points to predicted cost as a downside: http://semiaccurate.com/2015/07/29/i...ype-3d-xpoint/

Think how much 500GB of ram costs compared to a 500GB SSD, aim in the middle.

[KeyboardDemon]

Not quite, it is faster and possibly more energy efficient. If it was cheaper, I don't think Intel would be interested

The article, as I read it, suggests to me that reducing the cost of production is one of the main goals in this project, take a look at this quote:

The good news doesn't stop there as Intel and Micron scientists created the new chips to offer 1,000 times the endurance of NAND, lower power usage and to be more economical to produce.

In my mind the bit where they are looking for a more economical to produce technology is the bit that suggests they are looking for something that's cheaper to make. Assuming that this technology will supersede SSD, then I think it would be fair to say that they are looking for the speed and endurance advantages at a lower cost than the previous technology. But as I've also said previously, I don't think that lower production costs will mean lower ticket prices at the checkout.

[RobotDoktoru]

1,000 times faster than NAND

Does that make it faster or slower than conventional DDR RAM?

500 times slower than RAM, But 1500+ times more capacity RAMs