Unverified50% confidenceFactExact time
When model size reaches about twice the physical memory, mmap-based inference schemes collapse due to a 'page-fault storm', with the operating system blindly scheduling pages and inference stuck on disk I/O.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
Unverified当堆使用量超过虚拟机分配上限（由设备厂商配置，通常为256MB–512MB）时，系统会抛出OutOfMemoryError（OOM）导致应用崩溃67% similarUnverifiedThe strategy of substituting disk for VRAM performs much worse for dense models, because dense models require reading all parameters at every layer, multiplying the I/O pressure.67% similarUnverifiedOOM, or Out of Memory, refers to a situation where the memory requested by a program exceeds the available system limit, causing the operating system to forcibly terminate the process.66% similarUnverifiedUsing Unified Memory and letting the operating system automatically move data can drag down inference efficiency due to frequent paging under long contexts.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1017027API
curl https://kongchang.com/api/v1/knowledge/claims/1017027MCP
get_claim(id=1017027)