Samsung was announced the first demonstration of memory computing based on MRAM (Magnetoresistive Random Access Memory).
In traditional computing architecture, data is stored on DRAM memory chips and executed on processors (CPUs). Yet, In-memory computing is a new computing paradigm that aims to perform both storage and processing of data in memory.
As a large amount of data stored in the memory network itself can be processed and data processing is performed in a highly parallel manner, energy consumption is substantially reduced.
In-memory computing, therefore, has emerged as one of the Promising technologies the next generation of low-power AI semiconductor chips.
Memory computing is similar to our brain in that computing also takes place within the network of biological memories, or synapses, the points where neurons touch.
In fact, although at the moment the computation performed by our MRAM network has a different purpose than the computation performed by the brain, this solid-state memory network could be used in the future as a platform to mimic the brain, modeling the connectivity of brain synapses. .
Nonvolatile memories, particularly resistive random access memory (RRAM) and phase shift random access memory (PRAM), have been actively used to demonstrate this type of computation.
Yet, so far it has been difficult to use MRAM — another type of non-volatile memory — despite advantages such as operating speed, endurance, and large-scale production. This difficulty is due to the low strength of MRAM, so it cannot take advantage of reduced power when used in standard in-memory computing architecture.
Samsung researchers provided a solution to this problem by developing an MRAM array chip that demonstrates in-memory computing.
The solution was tested running AI applications and provided 98% accuracy in classifying handwritten digits and 93% in detecting faces in images. Samsung claims that MRAM technology can be used for AI processing and to create highly energy-efficient AI chips.