AMD发射MI430X GPU和432GB HBM4,用于AI和超级计算,为全球系统提供节能动力。 AMD launches MI430X GPU with 432GB HBM4 for AI and supercomputing, powering global systems with energy efficiency.

AMD已经启动了Instinct MI430X GPU,这是AI和HPC工作量的高性能芯片,432GB HBM4内存和19.6TB/s带宽。 AMD has launched the Instinct MI430X GPU, a high-performance chip for AI and HPC workloads featuring 432GB of HBM4 memory and 19.6TB/s bandwidth. 它支持FP64精确性和先进的AI行动,使科学计算和AI能够在一个单一的平台上有效执行任务。 It supports FP64 precision and advanced AI operations, enabling efficient execution of both scientific computing and AI tasks on a single platform. MI430X将为即将到来的超级计算机提供动力,包括Oak Ridge国家实验室发现器和欧洲Alice Recoque的Alice Recoque,后者与AMD的EPYC CPUs以及来自HPE Cray和Eviden的系统一起建造。 The MI430X will power upcoming supercomputers including Discovery at Oak Ridge National Lab and Alice Recoque in Europe, built with AMD’s EPYC CPUs and systems from HPE Cray and Eviden. 它的设计是为了提高能源效率,它使用AMD的ROCm软件,支持PyTorrch、TensorFlow和JAX。 Designed for energy efficiency, it runs on AMD’s ROCm software, supporting PyTorch, TensorFlow, and JAX. 该发射推进了AMD的HPC路线图,支持全球努力建立开放的、主权的AI基础设施。 The launch advances AMD’s HPC roadmap and supports global efforts toward open, sovereign AI infrastructure.