4 articles
Nvidia's new Vera Rubin AI compute platform promises a tenfold increase in performance per watt, driven by an "extreme co-design" of six integrated chips. This innovation tackles the growing demands of complex AI models and aims to enhance efficiency and reduce costs for data centers.
Nvidia's new Vera Rubin system promises a tenfold increase in AI performance per watt, addressing critical energy demands. Featuring advanced liquid cooling and NVLink technology, it aims to overcome AI's growing power consumption bottleneck.
NVIDIA CEO Jensen Huang unveiled the Vera Rubin architecture, a significant leap in AI computing power. The new system delivers up to 5x peak inference performance and 3.5x peak training performance over its predecessor, driven by innovations in chip design, networking, and system integration, all while enhancing energy efficiency.
Nvidia's new Vera Rubin platform introduces a novel approach to memory architecture, challenging AMD's dominance in high-bandwidth memory for AI. The innovation aims to overcome key AI bottlenecks and could reshape the competitive landscape.
