IT之家 AI

SK Hynix launches 3D FeNAND technology targeting AI computing

IT Home, October 15 – According to Korean media outlet The Elec’s report today, SK Hynix has announced its research findings on 3D FeNAND achieving Analog Compute-in-Memory (A-CiM). The related research has been…

Image source · IT之家 AI

IT之家 On October 15, according to Korean media The Elec’s report today, SK Hynix has announced its research results on 3D FeNAND achieving Analog Compute-in-Memory (A-CiM), and the related research has been published in a IEEE paper.

According to the introduction, A-CiM can use simulated signals to perform multiplication and addition operations directly in the memory (IT Home note: MAC), reducing data transfer, and enhancing processing speed and efficiency.

3D FeNAND uses the polarization state of ferroelectric materials to store data and control current. In contrast, traditional 3D NAND uses a charge trap nitride (CTN) structure, capturing electrons in the nitride layer. The characteristic of ferroelectric materials is that their internal polarization state remains intact even when the external electric field disappears.

SK Hynix stated that FeNAND can generate multiple current states within a single storage cell, suitable for the weight data required by AI computing.

Original source

IT之家 AI

Content notes

Original publication and rights belong to the source.

Machine translation · Refer to the original