Samsung
HBF Arrives: Samsung's Bet to Reshape AI Inference Economics
HBF (High Bandwidth Flash) is a new NAND-based memory. Samsung bets it can cut AI inference costs — making large models cheaper to run.
Back to home
AI inference
3 articles tagged with this topic
Nydus
AI Cold Start Cut from 116s to 1.4s — A Hidden Inflection in Compute Bills
Scaling AI inference, new nodes waited 116s. Nydus + JuiceFS cuts it to 1.4s and unifies model weights—quietly reshaping AI cost and elasticity.
C++20
C++20 Double Buffering Ends Data Queuing: Underlying Engineering Sets AI Limits
C++20 lock-free double buffering doubles memory to parallelize data generation and processing. As LLMs surge, it eliminates idle cycles caused by data