09/12/2024, 01:27 PM UTC
甲骨文云基础设施推出2.4泽塔FLOPS的AI超级计算机Oracle offering 2.4zettaFLOPS
➀ 甲骨文云基础设施（OCI）向客户提供云中最大的AI超级计算机，配备多达131,072个NVIDIA Blackwell GPU，提供2.4泽塔FLOPS的峰值性能；➁ OCI超级集群支持H100 GPU可扩展至16,384个GPU，提供高达65艾恩FLOPS的性能，而H200 GPU可扩展至65,536个GPU，提供高达260艾恩FLOPS的性能；➂ NVIDIA GB200 NVL72液冷实例使72个Blackwell GPU能够在单个NVLink域中以129.6 TB/s的总带宽相互通信。➀ Oracle Cloud Infrastructure (OCI) offers customers access to the largest AI supercomputer in the cloud, with up to 131,072 NVIDIA Blackwell GPUs and 2.4 zettaFLOPS peak performance; ➁ OCI Superclusters can scale up to 16,384 GPUs with up to 65 ExaFLOPS of performance for H100 GPUs, and 65,536 GPUs with up to 260 ExaFLOPS for H200 GPUs; ➂ NVIDIA GB200 NVL72 liquid-cooled instances enable 72 Blackwell GPUs to communicate at a bandwidth of 129.6 TB/s.
---
本文由大语言模型（LLM）生成，旨在为读者提供半导体新闻内容的知识扩展（Beta）。
