日本名古屋——日本汽车零部件供应商爱信周一表示，该公司利用磁性技术开发出一种新型半导体，其功耗比同类现有芯片降低约98%，这一突破有望帮助更高效地处理海量数据。
NAGOYA, Japan -- Japanese automotive parts supplier Aisin has developed a new semiconductor using magnetic technology that consumes about 98% less power than similar existing chips, the company said Monday, a breakthrough expected to help process large amounts of data more efficiently.
该片上系统采用磁阻随机存取存储器（MRAM），后者作为一种节能存储方案备受关注。作为丰田汽车的供应商，爱信与位于仙台的东北大学合作开发了这款芯片，利用了后者在MRAM领域的专长。
The system-on-chip incorporates magnetoresistive random-access memory (MRAM), which is drawing interest as an energy-efficient memory option. Aisin, a Toyota Motor supplier, developed the chip with Tohoku University in Sendai, tapping its MRAM expertise.
MRAM是一种非易失性存储器，断电后仍能保留存储数据，且具备高速读写特性。台积电和三星电子等芯片制造商也在研发这项技术。
MRAM is a type of nonvolatile memory, meaning it retains stored data even when power is shut off, and it boasts high read and write speeds. Taiwan Semiconductor Manufacturing Co. and Samsung Electronics are among the chipmakers working with the technology.
人工智能和图像识别技术在汽车行业的应用范围日益扩大。将数据在设备端而非云端处理的所谓边缘AI，结合摄像头和传感器，需求正在上升。
Artificial intelligence and image recognition technologies are being adopted in an increasingly wider range of areas in the automotive industry. Demand for so-called edge AI, in which data is processed on a device rather than the cloud, combined with cameras and sensors, is on the rise.
但AI性能提升也意味着数据量激增。这给采用DRAM或固态存储的传统片上系统带来挑战，导致处理速度变慢、能效降低。
But higher AI performance also means larger data volumes. This creates challenges for conventional system-on-chip devices with DRAM or solid-state storage, slowing processing speed and reducing energy efficiency.
爱信表示，其新技术可提升数据密集型应用的效率。（爱信）MRAM可替代上述存储类型以实现更高效处理，将基于AI的目标检测等任务的处理时间缩短约90%，功耗降低约98%。爱信将探索MRAM在图像识别系统等领域的应用。
Aisin says its new technology can improve the efficiency of data-intensive applications. (Aisin) MRAM can replace those types of storage for more efficient processing, reducing processing times for tasks such as AI-based object detection by around 90% and cutting power consumption by about 98%. Aisin will explore applications for MRAM such as in image recognition systems.