phase_memory-thumb-550xauto-65694

A new memory system could change the whole computer market.

The age of flash memory might be nearing a close even before those sexy SSDs really hit the mainstream market, as IBM has just announced that they’ve figured out a way to make phase-change memory a commercial reality within five years. What’s in it for you? Well, how does accessing your data about a hundred times faster sound?

Phase-change memory (or PCM for short) works by using said phase-change material to store bits of data. When heat is applied to this material, it changes from having a crystalline structure to an amorphous (i.e. non-crystalline) structure. Each structure has a slightly different electrical resistance, and by reading those resistances, you can tell whether the PCM is holding a one bit or a zero bit. All this happens very, very fast, about 100 times faster than conventional flash memory. This puts PCM nearly on par with the DRAM in your computer, except that unlike DRAM, PCM can remember your data without being left on all the time, even over 10 million write cycles or more.

There are two big problems with phase-change memory. Problem one is that over time (months and years), electrons in the phase-change material get bored and start to wander around, messing with the electrical resistance and potentially corrupting your data. This is unavoidable, but IBM Research has just come up with a way to compensate for electron wandering on the software side, using something called modulation coding.

What’s modulation coding? I was afraid you were going to ask that. As IBM explains it, “the modulation coding technique is based on the fact that, on average, the relative order of programmed cells with different resistance levels does not change due to drift.” So there you go then.

The second problem with PCM is that it’s a power hog. This is close to being solved, however, thanks to the magic of carbon nanotubes, which should reduce PCM power consumption by a factor of 100.Or maybe 1000. Either way, that’s enough to make it commercially viable.

Look for phase-change memory to show up first in enterprise (and cloud) storage systems sometime before 2016, and the trusty old trickle-down effect should put this stuff in your hot little hands soon thereafter. To get into your cellphone, the carbon nanotube solution will probably need to be implemented to meet mobile device power requirements, but multiple teams are working on that, too.

IBM Research

via Tech Review