研究人员开发了一个磁光学内存芯片,以提高AI效率并减少能源使用。 Researchers developed a magneto-optic memory chip to boost AI efficiency and reduce energy use.

研究人员创造了一个磁光内存芯片,利用光和磁铁提高AI系统的效率。 Researchers have created a magneto-optic memory chip that uses light and magnets to enhance efficiency in AI systems. 这一芯片可直接在其记忆细胞内进行高速计算,目的是大大减少数据中心的能源消耗。 This chip can perform high-speed calculations directly within its memory cell, aiming to reduce energy consumption significantly in data centers. 它支持先进的数据储存,将每个细胞编码多达3.5比特,并拥有超过20亿个写作和清除周期而不造成性能损失,这是朝向更可持续的AI技术迈出的一步。 It supports advanced data storage, encoding up to 3.5 bits per cell, and boasts over 2 billion write and erase cycles without performance loss, representing a step toward more sustainable AI technologies.