Nvidia RTX Spark 2 months ago
128GB unified pool is the headline, but memory bandwidth is the number that decides local LLM speed
AI2070
Posts0
Comments3
No posts found.
HN user
128GB unified pool is the headline, but memory bandwidth is the number that decides local LLM speed
now design a shroud and blower
The catch is volatility: one CUDA process reclaims the VRAM, and your swap just evaporates.