Samsung announces first computer with MRAM non-volatile magnetoresistive memory: can be used for AI computing

According to foreign media techpowerup, Samsung Electronics today officially announced the world's first computer equipped with MRAM memory. The relevant paper was published on the Nature website and will be published in a magazine soon. The title of the paper is "Cross-array of magnetoresistive memory devices for in-memory computing computers", and the related computer can be used for AI computing.

The technology was developed jointly by Samsung SAIT Research Institute and Samsung Electronics Foundry Business and Semiconductor R&D Center. The first author of the paper is Dr. Seungchul Jung.

Most current computers use independent RAM memory and independent hard disk storage, but in order to improve computing efficiency, the industry has been developing non-volatile memory that can take into account the functions of both hard disk and memory, while helping to greatly reduce power consumption.

To achieve this goal, researchers have developed RRAM (resistance random access memory), PRAM (phase change random access memory), etc., and prototypes have already been produced. The MRAM (magnetoresistive non-volatile random access memory) used by Samsung has the advantages of high speed, durability, and easy mass production, but due to power consumption issues, it has not been officially unveiled before.

Samsung Electronics researchers have provided a solution through an innovative architecture. Specifically, the researchers have successfully developed a new MRAM array that replaces the current architecture through a new "resistance" and computing architecture.

IT Home learned that after AI computing performance testing, computers equipped with MRAM memory have an accuracy of 98% in recognizing handwritten numbers and 93% in facial recognition.

Researchers said that MRAM chips used in in-memory computing computers are very suitable for neural network calculations, because this computing architecture is similar to the brain's neural network.