Micron 3D X Point Conference Call

posted in: Notes | 0

Since 1947 there have been seven classes of fundamental memory, each have brought different attributes to the way we process, store and retrieve data

“It’s really amazing to think that in the 50, 65 plus years since the invention of the transistor back in 1947, we have really only had seven different classes of fundamental different types of memory in the industry. And each of those memory classes has really brought different attributes for the compute hierarchy or to the way we process data and way we retrieve, store and retrieve data.”

You’ve had read only memories, SRAMs and DRAM

“If you go way back in the beginning, you had early read-only memories that were relatively dense, but could be written once and really didn’t have the ability to feed data at a high speed to a processing unit. Overtime, we had SRAMs that were much faster but volatile, so the data came and went, you had to restore the data overtime. We had, one of my favorite memories, DRAM, which is a system main memory today in many-many applications and has a nice combination of speed and density but not non-volatility.”

NOR and NAND flash were introduced 25-30 years ago

“And when I joined the industry over 30 years ago in 1984 we had the first NOR Flash product into the market and that was interesting because it brought reprogrammable non-volatile memory in a way that could interface directly with a processing unit. And then most recently a NAND Flash, way back 2.5 decades ago.”

This is a new class of memory that is truly revolutionary

“when we talk about what we’re doing today, introducing a new class of memory, it truly is a revolutionary and very-very exciting because it’s bringing not only a new set of characteristics to the marketplace, but it’s going to enable a whole new universe of applications and memory architectures and compute architectures that I think are really going to change the world and way we think about what’s possible in electronics.”

This is higher performance, very dense and non volatile

“What’s really exciting about the memory we are introducing today is that it’s higher performance, it can be very dense and it has non-volatile characteristics, data retention, endurance characteristics that Rob will illuminate in more detail here in a minute, that really allow us to do things that we haven’t been able to do before.”

This new technology is about getting the massive amount of data we’re collecting closer tot he processor

“So really a phenomenal amount of data being created in the world that’s really only useful if we can get it close to the processor and do something with it, either add it or subtract it or compare it to each other or use it in a way to really create useful knowledge as opposed to random information. And that’s what this new technology is all about, it’s about getting that data somewhere where we can get a lot of it, close to processing and into a useful format.”

This is something that many people thought was impossible

“This is something that many people thought, was impossible and many people gave up trying to accomplish. And it takes the power of companies like Micron and Intel and the sustained investment to work at something that other people thought was impossible.”

When you get this much memory close to the processor you can do things you couldn’t do before

“So as Rob points out the really exciting thing here is that when you can get this much memory close to a processor and you don’t have to refresh it because it’s non-volatile and it’s high performance and you can read it and write it many-many times, you really have the opportunity to do an amazing set of things that you couldn’t previously do”

Example: decrease loading time in video games

“every so often the character in the game might move to another scene and the game will stop and a little video runs — it reloads more data close to the processor. That is the fundamental bottleneck in terms of the amount of data you can get closer to the processor.”

If we had this technology when we created it we could have gotten to it faster

“I was joking with Rob in the back a little earlier about too bad we didn’t have this great technology when we were doing all the materials research to make these switches and memory elements work because had we had at it then we probably would have got there a little bit faster and so a really phenomenal high performance computing surge I think we’re going to see with this new memory as we get more memory close to the processor.”

This is a real technology, not a power point presentation

“What I was really trying to communicate to you is this is a real technology, it’s not a PowerPoint presentation. There is a real wafer here on the stage with us today. This is a manufacturable technology that Intel and Micron will be manufacturing first in our joint venture facility in Lehi, Utah. And beyond that, we will have to see where we go together. ”

The cost will be somewhere between NAND and DRAM

“you could put the cost somewhere between NAND and DRAM of course. So from a cost per bit, it’s likely to be in between those somewhere and a — but the actual costs will be resulting from the final products that we deliver to the marketplace in 2016 it is a very scalable technology”

It’s wicked fast storage

“this 3D XPoint technology is way faster, not a little bit faster it’s a 1,000 times faster than NAND technology and it’s writable in small amounts. So we can write words and so it can act more like memory as well as storage. But it’s like NAND and that is non-volatile and the data stays when the power goes off so you can use it as storage it’s just wicked fast storage.”

You have potential for innovation on the same level as when people moved from rotating hard disks to solid state drives

“the level of innovation that you saw there when folks moved from rotating hard disks to solid state drives in the enterprise and datacenters and cloud datacenters and in clients has the potential to happen again with a new technology.”

***It takes a tremendous amount of guts to work for an extended period of time to solve a problem that you don’t know is possible to solve***

“It’s one thing to pursue a problem that you know is solvable and whether that’s a next generation of something. It’s another thing to pursue a problem for an extended period of time that you don’t know is solvable, it might be impossible. And that — it requires commitment and investment, and vision and capability from these technology development teams that is hard to find. And I think that’s one of the unique things that has enabled this technology to come to life.”

You really can’t think of this as NAND or DRAM it will be used as both in different applications

” you shouldn’t think of this as NAND or DRAM, you should think of it as a whole new class of memory. It really does fill its own unique spot. Now it can be used in more of a storage type of application or it can be used more as a system main memory and we think it will be used as both for different applications and for different reasons, but it really kind of fits in that unique spot.”