Des articles pour ceux qui
livrent le stack
Textes originaux sur React, Node.js, TypeScript et AI — issus de la même pratique que notre logiciel opérateur. Corps des articles en anglais.
Tag : llm-inference
QUOKA Explained: Choosing Which KV History Attention Should Read
How QUOKA speeds up long-prompt prefill: KV cache basics, chunked prefill, picking outlier queries, max scoring, and cheap pre-aggregation for grouped-query attention.
2028 motsLireFlashAttention and PagedAttention: Two Fixes for Two Different Bottlenecks
Learn how FlashAttention speeds up attention math through tiling and online softmax, how PagedAttention manages KV cache memory, and why vLLM-style stacks use both.
1662 motsLireThe KV Cache Explained: Why Long Context Is a Memory Bill, Not Smarts
A plain-language guide to the KV cache: what it stores, how to compute its size, why long context eats memory, and four ways to cut the cost on local models.
1921 motsLireCerebras Wafer vs GPU for LLM Inference: A GPT-OSS Benchmark
Compare Cerebras' wafer-scale engine with GPU inference using GPT-OSS-120B across five workloads, then see how Ultrafast speeds up GPT-5.6 Sol through the API.
861 motsLireMoving KV Cache Between Qwen3 Models Instead of Re-Prefilling
NVIDIA research maps Qwen3-14B attention state into Qwen3-32B with a ridge-regression bridge. See what the latency numbers mean, where it fails, and why agents care.
1817 motsLire
À propos de ces articles
Devis en 24 h
Besoin du même stack en couche opérateur ? Envoyez le brief — devis sous 24 h.