Feed Entities Podcasts Deep Dives Developers About

[KongchangAI]

Real-time AI knowledge base. Every claim is traceable to its source, with data freshness under 24 hours.

Powered by 240+ first-hand AI sources

Knowledge

Feed
Hot Entities
Podcasts

Developers

CLI
MCP Server
REST API
About

© 2026 KongchangAI. The freshest, most accurate AI knowledge

Unverified50% confidenceFactExact time

Decode阶段属于访存密集型操作，整体吞吐受制于显存带宽而非算力

1

Sources

50%

Confidence

Long-term

Relevance

7/22/2026

First Seen

Sources

V100显卡选购对决：4×16G PCIe与2×32G SXM哪个更值？

Related Claims

UnverifiedPrefill阶段是计算密集型操作，Decode阶段是显存带宽密集型操作83% similar UnverifiedDecode 阶段属于访存密集型操作，每生成一个新 token 需从显存读取全部历史 token 的 KV Cache71% similar UnverifiedDisaggregated Inference架构将Prefill和Decode两个阶段分配到不同的硬件实例上执行70% similar Unverified受限解码（Constrained Decoding）在自回归生成的每个Token步骤上，根据有限状态机或上下文无关文法计算合法Token集合并重新归一化概率，从数学上保证输出符合目标schema68% similar Unverified编码器-解码器架构相比纯解码器架构，编码器一次性压缩长输入为固定维度语义向量，解码器专注生成短结构化输出，是Needle针对工具调用任务的工程取舍68% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/586049

API

curl https://kongchang.com/api/v1/knowledge/claims/586049

MCP

get_claim(id=586049)