New type of memory may make HDDs and SSDs obsolete!

Look up "racetrack memory" on Google News.

Basically, I'll sum up: You know how hard drives and solid-state drives both use magnetic fields to write to a disk? Over time, the wear and tear of the magnetic fields causes the disk to slowly break down, but not only that, the speed isn't as good as it could be.

There's several attempts to develop the next "big thing" in storage space, but one of them is coming along quite nicely, and it's called racetrack memory.

Basically, small electrons travel through extremely tiny wires in an endless loop at speeds of up to 100 MPH, and when needed, their movement is instantly stopped. It can stop their movement, read them, then resume movement, allowing for read/write speeds that are over a hundred times faster than the current magnetic disk model. It also results in no wear and tear, at least so far as they can tell.

The cost to manufacture is apparently low enough that it may replace the hard drives and solid-state drives we use today. And they can also fit terabytes of info into all this. So the result will be cheap, superfast drives with tons of storage space.

Just imagine having your next smartphone or tablet hold 1 TB of space instead of 64 GB (what I have on my current iPod). It'll be awesome. Space won't even be any issue, and you'll never think "man, I need to delete stuff again?!"

I'm pretty excited about this. Makes me wonder what the next quantum leap in processing speed will be as well.

http://www.eetimes.com/electronics-news/4076910/-Missing-link-memristor-created-Rewrite-the-textbooks-

That is all. Just to give you an idea, here's the quote:

HP has already tested the material in its ultra-high-density crossbar switches, which use nanowires to pack a record 100 Gbits onto a single die

I can't find it mentioned in the article, but the material can apparently by layered about 1000 times. The final device has a capacity of 100TB/cm³, and a speed comparable to that of DDR 2 RAM.

edited 14th Jan '11 7:21:24 AM by Yej

Uh, wow. I was wondering when memristors would get more prominent.

Space won't even be any issue, and you'll never think "man, I need to delete stuff again?!"

Every time my storage capacity gets exponentially larger, I think this, and then I end up filling it anyway.

There's multiple attempts to replace the current model of memory and storage space. There's even a group trying to create a form of holographic memory, where all information is stored as light in 3D space that is then read.

The big question is cost and speed, along with storage space.

If memristors cost too much to manufacture, then a different form of storage will probably catch on instead. That's part of why racetrack memory is probably a big deal now: according to its inventor, it can be cost-effective to manufacture.

Hmm an interesting practical application of quantum mechanics. Maybe they'll be ready for deployment in a few years but it does look like a nightmare to manufacture from what I read of its design.

Hard drives have relied on quantum physics for years.

Not at all. There's a difference to wrestling with problems arising due to quantum mechanics and finding ways around it compared to utilising it to your advantage.

Noob here: If it uses electrons as the data storage medium, does the block of memory need a power source...?

And some other throughts related to that, but my brain's "nebulous-thought" translator is acting up again. Anyone else get that problem - some magging thought that merely flits with consciousness, but it eludes any and all attempts to focus down onto it?

Let me repeat that for you: Memristors operate at nearly the same speed as RAM. Cost is almost irrelevant when you have 5 Libraries of Congress in a space the size of your thumb.

Noob here: If it uses electrons as the data storage medium, does the block of memory need a power source...?

Well, memristors are called memristors because they «remember» the charge that was passing through them. Or the voltage, or something, I suck at electronics, but I think that's the idea.

Yes, the problem with RAM is that it needs active current to maintain a memory state. Memristors, like NAND (Flash) memory, retain their state in the absence of power, but they allow far greater storage density.

The article said it remembers changes in the current passing through it, by changes to its resistance. Which makes sense, since changing resistances are what cause drops in voltage...or at least that's how I remember it. Of course I only ever took Physics 2, so I don't really know much.

I think it's funny that they've been doing all this workaround math with charges and voltage for almost 40 years, when what they were really looking for was the relationship between charges and flux, the derivative/rate of change of voltage.

edited 14th Jan '11 7:55:29 PM by deathjavu

Noob here: If it uses electrons as the data storage medium, does the block of memory need a power source...?

And some other throughts related to that, but my brain's "nebulous-thought" translator is acting up again. Anyone else get that problem - some magging thought that merely flits with consciousness, but it eludes any and all attempts to focus down onto it?

It requires power to change its value but not to maintain it, hence it being non-volatile, ohterwise turning off your computer wipes all your memory. Most new memory schemes today attempt to utilise quantum mechanical effects to its advantage. In comparison, our current HD Ds usually work around them (such as spontaneous bit flips due to thermal radiation).

HOLY FUCK

This is AWESOME

I can't wait till I have one in my computer. Never again will I worry about the size of my R Factor folder!

Not at all. There's a difference to wrestling with problems arising due to quantum mechanics and finding ways around it compared to utilising it to your advantage.

http://en.wikipedia.org/wiki/Giant_magnetoresistance

^ I was actually going to use that same thing for my argument

The point is that they're using the property specifically to counter the issue arising out of miniaturisation of something not intended to be significantly affected by random thermal bit flips. They are counteracting problems now arising due to the bits being so small hence the higher stability of drives back in the day.

It's same as the problem of our current CMOS transistors leaking current. We've made them so small that the insignificant amount of current that slips through due to quantum tunnelling is actually causing heat and power issues now and we need to use some other quantum mechanical effect to counteract it.

I say go back to using tubes. Doubles as a heating source for your home and even a nice lightsource, too.

Especially when it explodes into flames!

I just don't know which of the new memory technologies is the one going to take off. They have so many going on right now. What they're all demanding is ridiculously high-precision manufacturing, which is extremely costly. Whoever can design something that is really simple to build is going to really take the cake no matter how much more efficient these other types are. Afterall, our CP Us are built off transistors which are far less powerful than an earlier design simply because these wouldn't require a nuclear reactor to play doom 3.

edited 14th Jan '11 12:46:09 PM by breadloaf

Got a guuitar player in our church that will NOT play off a transistorized amplifier. Yeah, he's a little wonko.

But yes, the simpler process - the 80-percent solution - will win over the 100-percent solution that is a total pain to produce.

Well, at that scale, everything requires high-precision manufacturing. After all, the transistors themselves are smaller than light itself.

Umm, I really don't think you mean that. Smaller than the wavelengths of visible light, maybe, but photons are massless particles. They don't technically have a size.

Well, they're also waves

Waves come in many lengths, and calling it a "size" is a bit of a case of You Fail Terminology Forever. It is true that circuit design is currently gauged in terms of a certain number of nanometers resolution, which reflects the wavelength of light used to etch the traces on the boards. I understand that they're using some rather fancy tricks these days to design circuits that are actually smaller than the wavelengths of the lasers in question, although I don't pretend to be an expert on these matters.

Wave vs. particle, and there's a few people that honestly think the universe is a simulation, becasue of how goofy light behaves.

...which... is quite a neat proposition to consider.

The Truman Show, but on a massive scale?