11-19 更新 实时
首页/ 算力/ 正文

NVIDIA DSX MaxLPS:AI 工厂吞吐最大化

算力 来源:NVIDIA Developer 2026-11-07 02:40 0 阅读 机器翻译
NVIDIA DSX MaxLPS:AI 工厂吞吐最大化

每一瓦未使用的电力,都是被白白闲置的容量。AI 工厂通常是为每个 GPU 都达到峰值功耗这种不太可能出现的时刻进行配置的,这就形成了一层保护性缓冲,却可能在正常运行期间导致宝贵的基础设施未得到充分利用。NVIDIA DSX MaxLPS 使用策略管控的电力共享,在参与资源之间动态分配电力,使客户能够部署最多……

译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。

查看英文原文(How NVIDIA DSX MaxLPS Maximizes AI Factory Throughput and Efficiency)
Server rack connected to a power gauge and control nodes, illustrating AI factory power management with NVIDIA DSX MaxLPS.Every unused watt is capacity left on the table. AI factories are typically provisioned for the unlikely moment when every GPU reaches peak power, creating a...Server rack connected to a power gauge and control nodes, illustrating AI factory power management with NVIDIA DSX MaxLPS.

Every unused watt is capacity left on the table. AI factories are typically provisioned for the unlikely moment when every GPU reaches peak power, creating a protective buffer that can leave valuable infrastructure underused during normal operation. NVIDIA DSX MaxLPS uses policy-governed power sharing to allocate power across participating resources dynamically, enabling customers to deploy up to…

Source

正文保留英文原文,标题与摘要由机器翻译。

本文由采集器自动同步

原文发布于 NVIDIA Developer(How NVIDIA DSX MaxLPS Maximizes AI Factory Throughput and Efficiency),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。

阅读原文 ↗