German researchers devise means for even faster MRAM

We've already heard plenty of claims about how MRAM (or magnetoresistive RAM) would leave traditional memory in the dust, but a group of German researchers are now saying that they've found a way to blow past the MRAM that IBM and others are just now putting the finishing touches on. Their big breakthrough, it seems, is a new spin-torque system that can flip the magnetic field storing the 1s and 0s with only a single "wobble" being allowed to take place. That's as opposed to the magnetic field in current MRAM systems, which requires some time to settle into place each time its flipped. There's apparently still some work to be done before we see any actual MRAM using this new system, however, as the current used by the German device too electrically dense to be supplied by the transistors used in MRAM circuits, although they seem confident that they can overcome that pesky little problem.

[Flashpoint @ Aug 13th 2008 1:55PM]

What is the fastest theoretical transfer rate we can get from SSD?

What is the most likely transfer rate we'll be getting once the technology matures?

since SSD's are solid state and so are RAM DIMMS, would the new SSD's benefit from redesigned interfaces with the RAM? Furthermore, would it make sense to redesign the connections of video cards and sound cards while making SSD's the new standard?

[neofolklore @ Aug 14th 2008 7:46AM]

long comment short: no.

[LarryLarryLarry @ Aug 13th 2008 2:05PM]

"current is too electrically dense"

I had been relying on mainstream wire services for meaningless parody of science reporting. Thanks for filling that gap, Engadget!

[patsy @ Aug 13th 2008 3:30PM]

In addition to that, you gave them the benefit of a verb where there was none. For shame!

[Matthew Hilario @ Aug 13th 2008 2:20PM]

So it's like Tesla's indcution motor for memory?

[KrustyOrama @ Aug 13th 2008 2:22PM]

They got "perpendicular"

[Krush @ Aug 13th 2008 2:47PM]

ze Germans ...

[corey.coughlin @ Aug 13th 2008 2:35PM]

It sounds like the current required is just higher than they can currently supply. Should be easy to fix by just sizing up the driver transistor, in theory. In practice, it's not entirely clear. They don't give a particularly good sense of what removes the wobble in this design, so if it's a specific current limit that they require, then they might have trouble getting it to work across the full range of process yields. But, it's all pretty vague with the science reporting here.

[uskali @ Aug 13th 2008 3:30PM]

That magnesium oxide layer is supposed to be 1nm. One magnesium atom is 0.15nm diameter. I can't imagine they'd be able to hold tolerances less than 10% considering that's.. ONE FUCKING ATOM

[LarryLarryLarry @ Aug 13th 2008 4:10PM]

Worse than that, there's supposed to be a layer of Cobalt/Iron/Boron molecules which is 1 nm thick. I'm sure if can be built with a scanning tunneling electron microscope force probe. I'm also sure it can't be done with any bulk manufacturing process on the planet.

[Ian @ Aug 13th 2008 5:10PM]

goddamn technology SLOW DOWN!! or at lest make it faster for all this kool shit to come to the consumer!

[Teron @ Aug 13th 2008 6:06PM]

bastard Germans always stealing my ideas..

[neofolklore @ Aug 14th 2008 7:47AM]

BSOD waiting to happen.