Chinese AI team wins global award for replacing Nvidia GPU with industrial chip
Chinese researchers have developed an AI video-generation model, FlightVGM, that outperforms Nvidia's high-end GPUs in speed and energy efficiency while running on a widely available AMD FPGA chip. This marks a significant challenge to Nvidia's dominance in AI hardware and signals a potential shift in the global AI hardware race.
FlightVGM achieved a 30% performance boost and 4.5 times greater energy efficiency than Nvidia's RTX 3090 GPU. The model, developed by researchers from Shanghai Jiao Tong University, Tsinghua University, and Infinigence-AI, earned the Best Paper Award at the FPGA 2024 conference. This innovation could redefine the deployment of cost-effective and energy-efficient AI systems across various industries, including robotics and autonomous vehicles. FPGAs, unlike CPUs and GPUs, offer customizable architecture, leading to higher energy efficiency, while Nvidia GPUs are known for their parallel computing power.
via SCMP Full Text Feed
Chinese researchers have developed an AI video-generation model, FlightVGM, that outperforms Nvidia's high-end GPUs in speed and energy efficiency while running on a widely available AMD FPGA chip. This marks a significant challenge to Nvidia's dominance in AI hardware and signals a potential shift in the global AI hardware race.
FlightVGM achieved a 30% performance boost and 4.5 times greater energy efficiency than Nvidia's RTX 3090 GPU. The model, developed by researchers from Shanghai Jiao Tong University, Tsinghua University, and Infinigence-AI, earned the Best Paper Award at the FPGA 2024 conference. This innovation could redefine the deployment of cost-effective and energy-efficient AI systems across various industries, including robotics and autonomous vehicles. FPGAs, unlike CPUs and GPUs, offer customizable architecture, leading to higher energy efficiency, while Nvidia GPUs are known for their parallel computing power.
via SCMP Full Text Feed
