[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-5f745fe5-ea5d-453a-8b08-7dac524d1ac2":3},{"id":4,"title":5,"summary":6,"original_url":7,"source_id":8,"tags":9,"published_at":23,"created_at":24,"modified_at":25,"is_published":26,"publish_type":27,"image_url":13,"view_count":28},"5f745fe5-ea5d-453a-8b08-7dac524d1ac2","ACL 2026 Findings 把 KV 缓存优化重塑成「系统学」:综述 sKis 用三层视角搭起 LLM serving 坐标系","近两年关于 KV Cache 压缩、量化、驱逐、prefill\u002Fdecode 拆分的工作呈爆炸式增长,但大多在孤立 benchmark 上自报收益。ACL 2026 Findings 收录的综述「Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization」(arXiv:2607.08057)第一次把这场「算法竞赛」拉回系统工程语境。\n\n论文核心是 sKis 框架(system-aware KV infrastructure for serving LLMs),把现有方法归入三个维度:时间轴覆盖调度、流水线与硬件感知执行;空间轴把布局与迁移分为 GPU 内存层级和跨计算设备两层;结构轴则涵盖量化、低秩近似、结构压缩、驱逐策略与生命周期管理(KVCC \u002F KVRM)。\n\n更有价值的是随附的 behavior × objective 矩阵——表格显式标注每个方法主要改善的是平均延迟、长尾延迟、吞吐、显存还是互联 I\u002FO,并把「质量损失」作为独立维度列入。研究指出,≥70% 的现有论文只报告其中两项收益,对互联争用、能耗、quality impact 几乎不提及。\n\n对工程团队,sKis 提供了「按瓶颈选技术」的判别工具:部署卡在显存时翻 §5.1,卡在跨卡带宽时回到 §4.2。下一篇 KV 论文若不在这张矩阵里写明自己落点,视野就明显窄了。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.08057","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"2d9c2fb0-2be5-4ad1-aedb-e9747addf355","compression",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":18,"name":19,"slug":19,"description":13,"color":13},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",{"id":21,"name":22,"slug":22,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm","2026-07-12T18:15:00Z","2026-07-12T18:12:39.838976Z","2026-07-12T18:12:39.838993Z",true,"agent",6]