Unverified50% confidenceFactExact time
num_ctx 参数直接影响内存占用，KV缓存需要为每个上下文token存储注意力计算的中间状态
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
Building a Local AI Coding Agent on Mac: A Complete Guide with Ollama, OpenCode, and MCP
redditr/ollama7/11/2026
Related Claims
Unverified记忆在LLM系统中会在显式存储层、摘要层、上下文缓存层和间接推理层等多个层面留下痕迹，形成记忆残影71% similarVerifiedMulti-head Latent Attention (MLA) compresses storage by projecting KV into a low-dimensional latent space, reducing KV Cache memory usage to 5% to 13% of that of traditional methods.67% similarUnverifiedTraditional KV cache implementations pre-allocate contiguous GPU memory blocks of maximum length, resulting in an average of 60%–80% of pre-allocated memory being wasted.65% similarUnverifiedNumPy默认采用行优先（C-order）内存存储65% similarUnverifiedIn vLLM, each memory page typically stores KV vectors for a fixed number of tokens (e.g., 16), allowing KV Cache from different requests to be stored non-contiguously in physical memory.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/490101API
curl https://kongchang.com/api/v1/knowledge/claims/490101MCP
get_claim(id=490101)