|From left are Harvard University professor and Samsung Advanced Institute of Technology (SAIT) fellow Ham Don-hee, SAIT staff researcher Jung Seung-chul and SAIT Vice President of Technology Kim Sang-joon. Samsung Electronics said Thursday its researchers demonstrated the industry's first magnetoresistive random access memory-based in-memory computing technology. Courtesy of Samsung Electronics|
By Baek Byung-yeul
Researchers of Samsung Electronics successfully demonstrated magnetoresistive random access memory (MRAM)-based in-memory computing that simultaneously stores and processes data for the first time in the world, which could be utilized later to produce low-energy artificial intelligence (AI) chips, the semiconductor giant said Thursday.
In-memory computing has been regarded as a next-generation AI chip technology, as it performs both data storage and processing in a memory network, in contrast to the standard computer architecture in which data is stored in memory chips and processed in separate processor chips. As the data processing in the memory network is executed in a highly parallel manner, it reduces energy consumption.
Samsung said its researchers at Samsung Advanced Institute of Technology (SAIT) successfully demonstrated the technology using a new type of non-volatile memory chip, called MRAM.
Their innovative research result was published online by science magazine Nature on Jan. 12, and is set to be published in the upcoming print edition of the magazine. Jung Seung-chul, a staff researcher at Samsung Advanced Institute of Technology (SAIT) and the first author of the paper, and co-corresponding authors ― SAIT fellow Ham Don-hee and SAIT Vice President of Technology Kim Sang-joon ― spearheaded the research.
"In-memory computing draws similarities to the brain in the sense that in the brain, computing also occurs within the network of biological memory, or synapses, the points where neurons touch one another," said Jung, one of the paper's first authors, adding that their demonstration will help with R&D on neuromorphic technology that mimics the human brain.