IBM researchers are reaching beyond the perceived limits of memory technology

article image

Tuesday, Mar 22, 2016

Computing memory has evolved a great deal over the years, from drums to tubes to random access memory (RAM) to dynamic random access memory (DRAM) to flash—and that’s the short list. This is indicative as to how computing expectations (cheaper, smaller, faster) push researchers to reach beyond the bleeding edge.

Computing memory has evolved a great deal over the years, from drums to tubes to random access memory (RAM) to dynamic random access memory (DRAM) to flash—and that’s the short list. This is indicative as to how computing expectations (cheaper, smaller, faster) push researchers to reach beyond the bleeding edge.
This is also the case within IBM because some of today’s emerging workloads, such as in-memory analytics and computing, may not be ideally served by current memory technologies. However, explains Evangelos Eleftheriou, IBM Fellow and head of the cloud and computing infrastructure department at the tech giant’s Swiss lab, alternatives on the horizon will continue to help enterprises meet data-processing needs, even if some may seem straight out of science fiction.
“We foresee that PCM can be used as a hybrid system together with flash. In this way, we can solve some of the problems with the write endurance of flash.”
—Evangelos Eleftheriou, IBM Fellow