Semiconductor manufacturers are speeding up the development of neuromorphic which combines the memory semiconductor and processor to calculate like human brains. It produces high performances with low power. So, neuromorphic semiconductors are emerging as an alternative to process rapidly increasing data. This will be widely used in AI, IT, and autonomous vehicles.
Today's semiconductors are connected in series and are used. Memory semiconductor and CPU carry out each mission separately. The performance of exchanging data by connecting the two semiconductors has not changed much. However, neuromorphic semiconductors can provide data to each other. Neuromorphic semiconductors are connected in parallel, so they can calculate and store the data at once like human brains. In case of neuromorphic semiconductors, the degree of agglomeration is 3,500 times higher than ordinary semiconductors. So, the computing power is thousands of times faster than ordinary semiconductors.
In South Korea, Samsung and SK Hynix speed up the development of neuromorphic semiconductors. Samsung chooses this as a future skill. It is known to have been showing 2.5 times faster in processing and a 60 percent low in power consumption in 2021. They also published a thesis in Nature Electronics that presented the vision of neuromorphic semiconductors. Samsung Electronics recently implemented the world's first in-memory computing based on magnetoresistive memory, which combines the advantages of DRAM and NAND flash memory. In-memory computing is a technology that performs data operation and storage in parallel with memory. Samsung Electronics believes that in-memory computing research will help develop neuromorphic semiconductors.
The prospect of neuromorphic semiconductors is bright in that field. Many semiconductor manufacturers give an effort to develop this skill, and it will be widely used in AI, IT, and autonomous vehicles in the future.
By Um Ki Yoon
다른 곳에 퍼가실 때는 아래 고유 링크 주소를 출처로 사용해주세요.
울산 문수중학교 1학년