Breakthrough featured by illustrious journal Nature
Samsung Electronics announced this week that it had developed a technology geared toward increasing the performance of chipsets suitable for next-generation AI chips.
The technology is called in-memory computing, which merges memory and system semiconductors to reduce power consumption and latency.
In particular, the Seoul-based company worked on in-memory computing with MRAM, a non-volatile memory, which researchers had struggled to deal with.
Samsung said that MRAM has many merits like operation speed, endurance, and large-scale production. But the low resistance of MRAM generated many difficulties.
“The Samsung Electronics researchers have provided a solution to this issue by an architectural innovation,” a Samsung official said.
“The chip achieved an accuracy of 98 percent in the classification of hand-written digits and a 93 percent accuracy in detecting faces from scenes.”
Samsung said that the breakthrough would be featured at the next edition of the peer-reviewed journal Nature, which already published it online.
A few Samsung researchers have led the development, including Jung Seung-chul, the first author of the Nature paper, and Prof. Ham Don-hee, who also works as a Samsung fellow.
“In-memory computing draws similarity to the brain in the sense that in the brain, computing also occurs within the network of biological memories, or synapses, the points where neurons touch one another,” Jung said.
“In fact, while the computing performed by our MRAM network for now has a different purpose from the computing performed by the brain, such solid-state memory network may in the future be used as a platform to mimic the brain by modeling the brain’s synapse connectivity.”