To start with some recent history first.

Back in the early 1990's a fellow by the name of Stuart S. P. Parkin working for IBM did something we can all appreciate. He took his observations of some odd magnetic quantum effects in semiconductors and spun it into higher density HDD read/write heads. It did take him 2 year, but the results were astounding. We all owe our iPods to Mr. Parkin.

Oddly enough though, what he's proposing now could make hard drives obsolete someday. A new type of non-volatile memory that is both faster than current Flash and DRAM with 10 to 100 times the capacity. And perhaps most notable aspect of race track memory being that of its simple approach.

A wire loop to magneticly charge the two tracks below.

The physics/engineering behind the new memory involves constructing a track of verticlly standing wire loops around the edge of a silicon chip. Electric current passed through the loops is then used to manipulate the magnetic charge of nano-scale magnetic regions (or wire tracks) below. It's somewhat akin to how a hard drive writes data to the platter, but on a smaller scale and without a spinning motor or twitching armature. In this case, the data medium does't move while the read/write heads do.

It does bode well in terms of future computin, this magnetic-racetrack memory. The memory would be non-volitaile, would require little current for the sub-molucular positioning of the read/write head(s). During their experimantations, Parkin and his team were able to slide a microscopic magnet along a notched wire at a speed of 100 meters per second. Their latencies accheived were even more impressive, with read/write times down at a nanosecond.

It's a small start, but future models could rely upon 100 coils, all moving in tandum to simutanioulsy read and write 100 clusters of memory. In fact, it could almost be compared to those days of yerteryear tape drives. In this case however, it's the tape that's stationary and the read/write head that moves.

“Finally, after all these years, we’re reaching fundamental physics limits,” Parkin said. “Racetrack says we’re going to break those scaling rules by going into the third dimension.”

Certianlly sounds impressive, but don't get too excited. What experiments Parkin and his team have compleated won't function as memory yet, they're simply the indvidual portions of the greater model. As for those involved and those observing, the usual tech industry humility applies.

There have been the expected suggestions of wrist-watch televisions, terabyte MP3 players and 100% solid state computers. There may even come the possibility of PID (Processing In Memory) for small scale computing.