Samsung demonstrates the world’s first in-memory computing based on MRAM

by ddos ·

Samsung recently announced the world’s first in-memory computing based on MRAM (Magnetoresistive Random Access Memory). Related papers are also published online by Nature. The paper, “A crossbar array of magnetoresistive memory devices for in-memory computing,” showcases Samsung’s leadership in memory technology and what it has done to merge memory and semiconductor systems in next-generation artificial intelligence (AI) chips.

The research was led by Samsung Advanced Institute of Technology (SAIT) in close collaboration with Samsung Electronics Foundry Business and Semiconductor R&D Center.  The team was led by  Dr. Seungchul Jung, Staff Researcher at SAIT, and the co-corresponding authors Dr. Donhee Ham, Fellow of SAIT and Professor of Harvard University, and Dr. Sang Joon Kim, Vice President of Technology at SAIT.

Samsung stop Exynos processors

“Samsung Exynos Packaging” by minimalist ㅤ, Wonchan Lee is licensed under CC BY-NC-ND 4.0

In modern standard computer architecture, data is stored in memory chips and data computations are performed in processors. In contrast, in-memory computing is a new way of computing designed to perform both data storage and data computation in an in-memory network. Since this solution can process a large amount of data stored in the memory itself without data transmission, and the data processing in the memory will be performed in a highly parallel manner, power consumption will be greatly reduced. Therefore, in-memory computing has become one of the most promising technologies for realizing the next generation of low-power artificial intelligence semiconductor chips.

Samsung researchers provided a solution through architectural innovation, successfully developing an MRAM array chip that “demonstrates in-memory computing, by replacing the standard, ‘current-sum’ in-memory computing architecture with a new, ‘resistance sum’ in-memory computing architecture, which addresses the problem of small resistances of individual MRAM devices.”
Samsung believes that MRAM chips can also be used as a platform for downloading biological neuron networks in the future because their computing architecture is more similar to the brain neuron network.

Tags: MRAMSamsung