Samsung becomes the first brand in the world to demonstrate MRAM technology for in-memory computing, putting it at the forefront of a highly innovative technology race.
Samsung, the world’s largest memory chip maker, has announced that it has become the world’s first to demonstrate MRAM (Magnetoresistive Random Access Memory) technology for in-memory computing.
The research was the result of collaboration between SAIT (Samsung Advanced Institute of Technology), Samsung Foundry and Samsung Semiconductor R&D. And the importance of this milestone is such that The study has been published in the prestigious scientific journal Nature.
Normally, the data is stored in memory chips (DRAM) and is processed by a CPU or an AP (Application Processor). But nevertheless, data storage and computing are done on the same chip with in-memory computing.
Since it is not necessary to transfer data from memory to processor and vice versa, a lot of time is saved. Data processing within memory is carried out in parallel, which represents significant energy savings.
Samsung claims that MRAM technology will be ideal for fields such as artificial intelligence processing.
Other in-memory computing solutions such as PRAM (Phase-change Random Access Memory) and RRAM (Resistive Random Access Memory) have been demonstrated around the world, but MRAM had been difficult to test, until now.
Despite the non-volatile nature of MRAM, its high operating speed, and its endurance, it was difficult to use for in-memory computing due to its low endurance.. It was also impossible to use the energy efficiency advantage of MRAM in a standard memory computing architecture.
Now, Samsung has created a solution to the problem thanks to its innovative new architecture. The South Korean company has developed an MRAM matrix chip that uses resistance summation rather than standard current summation.
The new solution counteracts the problem of small resistances of individual MRAM devices. The company’s new MRAM computing tested by running AI applications, it was 98% accurate in sorting handwritten digits and 93% in the detection of faces in images.
Samsung claims its new technology can be used to make energy-efficient AI chips.