跳到正文
NVIDIA Technical Blog· Dan Feigin·· 2 天前AI 评分58

NVIDIA 开源 Snapshot,可在数秒内启动和恢复 GPU 负载

Start and Restore GPU Workloads in Seconds with NVIDIA Snapshot

AI 导读

NVIDIA 发布开源组件 NVIDIA Snapshot,作为 NVIDIA DSX AI Factory Platform 的一部分,让已初始化的 GPU 负载可在推理副本之间复用,从而加快启动与恢复。原文指出,LLM 负载初始化即冷启动可能耗时数分钟,当推理需求激增或副本故障时,空闲 GPU 无法立即转化为服务容量。

正文 · 原文

Green cube with a neural-network diagram, surrounded by circular arrows beneath descending data blocks on a purple background.NVIDIA Snapshot, a new open-source component of the NVIDIA DSX AI Factory Platform, makes initialized GPU workloads reusable across inference...Green cube with a neural-network diagram, surrounded by circular arrows beneath descending data blocks on a purple background.

NVIDIA Snapshot, a new open-source component of the NVIDIA DSX AI Factory Platform, makes initialized GPU workloads reusable across inference replicas—speeding startup and recovery. When inference demand spikes or a replica fails, available GPUs don’t translate immediately into serving capacity. Initializing an LLM workload—a process known as a cold start—can take minutes…

Source

来源:NVIDIA Technical Blog · developer.nvidia.com