“According to Phison’s internal testing, aiDAPTIV+ can accelerate inference response times by up to 10 times, as well as reduce power consumption and improve Time to First Token on notebook PCs. Obviously, the larger the model and the longer the context, the higher the gain, so the technology is especially relevant for Mixture of Experts models and agentic AI workloads.”- tomshardware.com
来源： 汤姆斯硬件网站
简体中文 
English 
繁體中文 
日本語 
Deutsch 
한국어