X

New IBM memory promises faster, higher-capacity devices

A new type of memory called Racetrack is finally off the drawing boards at IBM. Its development should pave the way for mobile and desktop devices that are faster, store more data, and chew up less power.

Lance Whitney Contributing Writer
Lance Whitney is a freelance technology writer and trainer and a former IT professional. He's written for Time, CNET, PCMag, and several other publications. He's the author of two tech books--one on Windows and another on LinkedIn.
Lance Whitney
4 min read

A new kind of memory from IBM Labs is promising to revolutionize how much data we can store and how fast we can access it on our mobile and desktop devices.

After spending six years as a theoretical concept, the memory, dubbed Racetrack, finally is a huge step closer to reality. Researchers at IBM have recently confirmed that their theories of the physics behind Racetrack are valid and can be used to develop and manufacture this new type of memory.

Images of a racetrack designed to pin domain walls at specific positions.
Images of a "racetrack" designed to pin domain walls at specific positions. IBM

This revolutionary type of memory could open up a whole new world for laptops, smartphones, and other mobile devices. Users would be able to store as much as 100 more times data on their portable gadgets, perhaps keeping as many as 500,000 songs or 3,500 full-length movies on one mobile device. And since Racetrack would use considerably less power, a single battery charge could power a device for weeks rather than days or hours.

The new memory is also expected to play a role in desktop computers and servers, allowing them to access more data much faster. In some ways, Racetrack combines the best elements of flash memory and magnetic storage and could prove to be the one technology to someday replace current RAM, Flash RAM, and even conventional disk drives.

How does Racetrack work? Unlike conventional memory, which needs to seek out the data it needs, Racetrack automatically moves the data to where it can be used. That serves to not only speed up data access but allow much more data to be stored in a smaller area.

The memory is so named because it moves the magnetic bits of data along thin, nanowire "racetracks," 1,000 times finer than a strand of hair.

The data itself is stored in magnetic regions known as domains. Using the spin of individual electrons, Racetrack memory can move these domains at hundreds of miles per hour and stop them at atomically precise spots along the nanowire, allowing huge amounts of information to be retrieved in less than a billionth of a second.

Scientists at IBM were the first to measure the time and distance involved in moving these domain walls, giving them a clearer understanding of how to control their movement, thus transforming Racetrack from research into reality.

A paper due to be published today in Science Magazine reveals more of the physics behind this new form of memory. CNET spoke today with Stuart Parkin, the head researcher on the Racetrack project, to explain how it works and how it might be used.

For the past several years, Parkin and other scientists at IBM Labs have been conducting research into the physics behind Racetrack, eventually proving that the memory is more than just a concept and is something that would work in the real world. Recent developments in the field of spintronics, which focuses on the spin of electrons, is what allows for the movement of the bits along the nanowires in Racetrack memory, explained Parkin. That opens up the ability to read and write more data very quickly at specific points along those wires.

Experimental racetrack made of a curved nanowire. Electrical contact lines, made out of gold, are used to apply current pulses to move the domain walls along the nanowire.
Experimental "racetrack" made of a curved nanowire. Electrical contact lines, made out of gold, are used to apply current pulses to move the domain walls along the nanowire. IBM

With Racetrack memory, manufacturers could build laptops, smartphones, and other devices with considerably more memory in the same or even smaller amount of space than currently required, according to Parkin. Racetrack would also require far less battery power and could be used to read and write data an infinite number of times. In contrast, Flash memory is a slow type of memory, uses a lot of power, and can only handle the writing of data so many times before it wears out.

"So the average user would see their iPhone much faster," Parkin said. "The battery would last much longer. And they would be able to do much more complex and powerful computations because this type of memory is capable of supporting ultra-fast manipulation of the data."

Racetrack memory can also be written to indefinitely, said Parkin. Since there's no intrinsic mechanism to wear out, Racetrack can manage much more complex calculations and functions that typically require a large number of data reads and writes.

Now that the physics have been proved and demonstrated, the next challenge for IBM is to begin building prototypes. But Racetrack is a whole new shift in the paradigm of memory architecture, and one that will require new physics and new materials. As such, the question for Parkin is one of demonstrating that billions of Racetracks can be integrated with and built onto large wafers, which will be required to make devices at a low enough cost.

"That's something that will take considerable investment," said Parkin. "And these kinds of steps can be quite time-consuming. Now that we are at the development phase, it's more a question of obtaining this significant investment to build the prototypes quickly."