Verified80% confidenceFactExact time
During large model inference, actual GPU utilization is often less than 10%.
4
Sources
80%
Confidence
Long-term
Relevance
7/8/2026
First Seen
Sources
Related Claims
VerifiedDuring the autoregressive decoding phase, GPU compute utilization is often less than 10%, making it a memory-intensive operation.85% similarUnverified单条推理通常只能占用GPU SM利用率的5%～15%，而合理的批次大小可将利用率提升至60%～90%76% similarUnverifiedPure CPU inference is typically 5–10 times slower than GPU inference.75% similarUnverified轻量版模型单次推理的算力消耗往往只有旗舰版本的十分之一甚至更低73% similarUnverified相较于用GPU执行AI推理，NPU的功耗通常只有前者的1/5到1/1073% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/216174API
curl https://kongchang.com/api/v1/knowledge/claims/216174MCP
get_claim(id=216174)