跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2611.25112 · v1

TripleKV: Predict, Prefetch, Protect -- KV Cache Management for AI Agents

首次提交 2026-11-19 03:50
版本更新 2026-11-19 03:50
雷达观测 2026-11-21 12:18
arXiv ↗PDF ↗

摘要

LLM agents alternate between inference and tool execution, repeatedly reusing their growing context across calls. Under high concurrency, their combined key-value (KV) cache working set can reach hundreds of GiB, far exceeding GPU memory capacity. As idle KV cache blocks are evicted to slower storage tiers, returning agents incur expensive retrievals that stall inference. To address this problem, we present \emph{TripleKV}, a three-tier KV cache management system that hides the delays of these retrievals. %exploits tool execution to hide these delays. In agentic workloads, the KV blocks needed by the next inference request are known in advance; only the time of their reuse is uncertain. TripleKV uses a tool-aware latency predictor to estimate when each tool call will complete and prefetches the required KV blocks from SSD into host DRAM ahead of time. Unlike GPU prefetching, this avoids reserving scarce GPU memory and interfering with concurrent inference requests. To further improve DRAM utilization, TripleKV eliminates duplication of GPU-resident blocks in DRAM, and introduces a victim-cache mechanism that refills blocks from SSD upon their eviction from GPU memory. Across data, web, and coding agents, TripleKV reduces the fraction of requests requiring critical-path SSD reads from 45--76\% to under 10\%. It also reduces P95 time to first token (TTFT) by up to 49\% compared with vLLM extended with DRAM and SSD KV-cache tiers, and achieves 1.5-7×\times lower P95 TTFT than recent agent-aware systems.

作者备注:13 pages, 23 figures

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

目前只收录了一个版本。收录下一版后,可以对照标题、作者、分类和摘要。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v1 · 2026-11-19 03:50 · 最新观测版本

TripleKV: Predict, Prefetch, Protect -- KV Cache Management for AI Agents

LLM agents alternate between inference and tool execution, repeatedly reusing their growing context across calls. Under high concurrency, their combined key-value (KV) cache working set can reach hundreds of GiB, far exceeding GPU memory capacity. As idle KV cache blocks are evicted to slower storage tiers, returning agents incur expensive retrievals that stall inference. To address this problem, we present \emph{TripleKV}, a three-tier KV cache management system that hides the delays of these retrievals. %exploits tool execution to hide these delays. In agentic workloads, the KV blocks needed by the next inference request are known in advance; only the time of their reuse is uncertain. TripleKV uses a tool-aware latency predictor to estimate when each tool call will complete and prefetches the required KV blocks from SSD into host DRAM ahead of time. Unlike GPU prefetching, this avoids reserving scarce GPU memory and interfering with concurrent inference requests. To further improve DRAM utilization, TripleKV eliminates duplication of GPU-resident blocks in DRAM, and introduces a victim-cache mechanism that refills blocks from SSD upon their eviction from GPU memory. Across data, web, and coding agents, TripleKV reduces the fraction of requests requiring critical-path SSD reads from 45--76\% to under 10\%. It also reduces P95 time to first token (TTFT) by up to 49\% compared with vLLM extended with DRAM and SSD KV-cache tiers, and achieves 1.5-7×\times lower P95 TTFT than recent agent-aware systems.

查看此版本解读 →查看此版本原文 ↗