Could PCM be faster than we thought?

1 min read

Industry and academia have been pursuing phase change memory (PCM) with varying levels of enthusiasm for five decades. And yet, despite a number of advances, the technology remains tantalisingly just out of reach.

PCM is being pursued for its potential to act as a ‘universal memory’ – something with the speed of DRAM and the non volatility of flash. But one of the stumbling blocks has been speed; PCM hasn’t yet matched DRAM’s ability to store data quickly. Although products have reached the market, their success has been limited.

However, work at Stanford University in the US suggests PCM might be faster than some think – in fact, up to three orders of magnitude quicker. What Stanford’s team has done is to focus on how long it takes for the crystalline version of a chalcegonide glass to turn into the amorphous state. Previously, it was thought this transistion took around 1ns, but Stanford’s team says it could actually happen in 1ps. If so, that could mean PCM is 1000 times faster than was thought. And if it is, will this bring PCM based products to market any faster?

Interestingly, Stanford’s work was published just ahead of a demonstration by Micron at the Flash Memory Summit of QuantX – its version of the 3D Xpoint technology developed in association with Intel. When it was unveiled last year, that generated more questions than both companies appeared to be prepared to answer. Since then, silence.

One of the questions was ‘just what is 3D Xpoint?’. Speculation since the launch suggested it could be a variant of PCM, but Micron and Intel described 3D Xpoint as a ‘new technology’ – and PCM isn’t new.

One thing which Micron has managed to do, however, is to take understatement to a new level. A blog on its website says ‘Micron has been fairly quiet on the technology’.