Hot Chips 2018: Nanotubes as DRAM from Nantero, a Live Blog

[Senpuu]

https://www.anandtech.com/show/13252/hot-chips-2018-nanotubes-as-dram-by-nantero-live-blog

The last I heard about this type of tech was sometime last year and my understanding was that it was going to be fabbed first party and rolled out on older, more accessible nodes and at very small capacities, breaking into the market as a niche product for industry. I work in the engineering field for industrial systems, so I was excited for this for many applications, but I didn't think it was nearly this close to mainstream. This news piece describes a very different story (IP house only, not fab; solid design and testing completed and ready for roll out; contract already made with Fujitsu, with market splash in 1-3 years).

What am I missing? Why isn't this a huge talking point here? This is better than anything available, both now or looking down the road a decade for any iterative improvement roadmap. And not just for RAM or for storage, but for both. It seems like it will be a game changer for the basic memory and storage paradigm that has been largely unchanged from a high-level perspective since the dawn of the PC.

 

[whm1974]

Companies have been working on using carbon nanotubes for computer memory and other chips for a long time now.

 

[Senpuu]

Obviously... but that doesn't address a thing I wrote.

 

[whm1974]

Obviously... but that doesn't address a thing I wrote.

Well to be honest, I haven't been keeping up with computer related carbon nanotube stuff for awhile now. I also thought nanotube memory and storage was suppose to be much faster then silicon based ones?

 

[maddie]

https://www.anandtech.com/show/13252/hot-chips-2018-nanotubes-as-dram-by-nantero-live-blog

The last I heard about this type of tech was sometime last year and my understanding was that it was going to be fabbed first party and rolled out on older, more accessible nodes and at very small capacities, breaking into the market as a niche product for industry. I work in the engineering field for industrial systems, so I was excited for this for many applications, but I didn't think it was nearly this close to mainstream. This news piece describes a very different story (IP house only, not fab; solid design and testing completed and ready for roll out; contract already made with Fujitsu, with market splash in 1-3 years).

What am I missing? Why isn't this a huge talking point here? This is better than anything available, both now or looking down the road a decade for any iterative improvement roadmap. And not just for RAM or for storage, but for both. It seems like it will be a game changer for the basic memory and storage paradigm that has been largely unchanged from a high-level perspective since the dawn of the PC.

I felt the same way when I first read about this, and I agree that this looks very big and rather ingenious. Basically an electro-mechanical device at its core.

 

[whm1974]

I felt the same way when I first read about this, and I agree that this looks very big and rather ingenious. Basically an electro-mechanical device at its core.

Yes but when will we start seeing memory and storage devices using this tech?

 

[IntelUser2000]

There's probably a sacrifice. Usually all the DRAM or faster than DRAM NV memory has been very low in density. The somewhat slower ones like 3D XPoint has higher density.

Some materials suggest retention rate is inversely proportional to performance among contenders. Meaning the slower the memory tech is, it lasts longer before data loss occurs.