Unverified80% confidenceFactTime unknown
一个拥有10,000块GPU的训练集群,假设单块GPU年故障率为3%,平均每天约有一块GPU出现问题
1
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDuring testing, only 1 of the 4 GPUs reached a utilization rate of approximately 10%–13%, while the rest were nearly idle, and CPU usage was as high as approximately 50%.71% similarUnverifiedAccording to surveys by organizations such as MLCommons, the average effective GPU utilization in large-scale industry training tasks is generally in the range of 40%–60%.69% similarUnverifiednvidia-smi报告的GPU利用率本质上只统计采样时间窗口内GPU上至少有一个内核在执行的时间占比,而非计算资源的实际占用率67% similarUnverifiedDuring large model inference, actual GPU utilization is often less than 10%.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/21231API
curl https://kongchang.com/api/v1/knowledge/claims/21231MCP
get_claim(id=21231)