https://www.youngju.dev/blog 천천히 올바르게. AI Researcher & DevOps Engineer Youngju's blog. GPU/CUDA, LLM, MLOps, Kubernetes AI workloads, and data engineering — plus mindset essays on confidence, routines, health, and sport psychology. ko fjvbn2003@gmail.com (Youngju Kim) fjvbn2003@gmail.com (Youngju Kim) Wed, 12 Aug 2026 00:00:00 GMT https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse.en https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse.en A rundown of when vLLM prefix caching hits and when it misses, based on the official design document. Covers how block hashes chain to the block before them, why only full blocks get cached, how a single timestamp placed at the front of a prompt wipes out the entire cache, and cache_salt. Part 5 of the Inside vLLM series. Wed, 12 Aug 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) vllmprefix-cachingkv-cacheprompt-engineeringllm https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse.ja https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse.ja vLLMのプレフィックスキャッシュがいつ効いていつ効かないかを、公式設計ドキュメントを基準に整理する。ブロックハッシュが直前のブロックに鎖のように連なる構造、フルに埋まったブロックだけがキャッシュされる理由、プロンプトの先頭に置いたタイムスタンプ一つがキャッシュ全体を無効にする過程、そしてcache_saltまで扱う。vLLM内部構造シリーズ第5回。 Wed, 12 Aug 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) vllmprefix-cachingkv-cacheprompt-engineeringllm https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse https://www.youngju.dev/blog/ai-platform/2026-08-12-vllm-prefix-caching-and-kv-reuse vLLM의 접두사 캐싱이 언제 적중하고 언제 못 하는지를 공식 설계 문서 기준으로 정리했습니다. 블록 해시가 앞 블록에 사슬처럼 묶이는 구조, 꽉 찬 블록만 캐시되는 이유, 프롬프트 맨 앞에 넣은 타임스탬프 하나가 캐시 전체를 날리는 과정, 그리고 cache_salt까지 다룹니다. vLLM 내부 구조 시리즈 5편. Wed, 12 Aug 2026 00:00:00 GMT fjvbn2003@gmail.com (Youngju Kim) vllmprefix-cachingkv-cacheprompt-engineeringllm