Verified65% confidenceFactTime unknown
大语言模型的推理过程本质上是内存带宽受限（memory-bandwidth bound）的任务，每生成一个Token都需要将模型的全部权重从存储中读取一遍
3
Sources
65%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe context of a large language model is essentially the total amount of information the model can process in a single inference, typically measured in the number of tokens.79% similarUnverified语言模型生成 token 本质上是内存带宽受限任务而非纯算力受限79% similarVerifiedLarge language model inference is divided into two phases, prefill and decode: prefill is compute-intensive and sensitive to GPU processing power, while decode is a memory-bandwidth-bound task.75% similarUnverifiedLarge language model text generation is an autoregressive serial process that produces only one token at a time, making it impossible to fully parallelize the output stage, with higher computational intensity than the input encoding stage.73% similarUnverifiedLarge language models form parametric memory by compressing and encoding training corpora into weight matrices, which is distributed rather than exact storage, and this is one of the reasons for hallucinations about brand information.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/38144API
curl https://kongchang.com/api/v1/knowledge/claims/38144MCP
get_claim(id=38144)