From 9ddabee6aecd50ce88636799edbdf52908de7516 Mon Sep 17 00:00:00 2001 From: Gahow Wang Date: Wed, 27 May 2026 11:02:29 +0800 Subject: [PATCH] Remove 'capped' references from MEETING.md and PAPER_OUTLINE.md prose MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Companion to the figure cleanup: prose in §3.1 was still quoting "capped 31.6% APC" as one of the failure-mode datapoints. Same reason as the figures — capped is a workload manipulation, not a policy, so it doesn't belong in the §3.1 routing-policy narrative. Co-Authored-By: Claude Opus 4.7 --- MEETING.md | 2 +- PAPER_OUTLINE.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/MEETING.md b/MEETING.md index e0fed0a..9eca272 100644 --- a/MEETING.md +++ b/MEETING.md @@ -39,7 +39,7 @@ L = Λ · N · W_turn(L) # agentic, T_human≈0 ![](figs/f4a_apc_loss.png) -LMetric 56.9%、load_only 54.1%、capped 31.6% APC，远低于 79.6% 上界。23pp 缺口直接来自跨 instance 路由丢的 intra-session hit。 +LMetric 56.9%、load_only 54.1% APC，远低于 79.6% 上界。23pp 缺口直接来自跨 instance 路由丢的 intra-session hit。 ### 静态 PD-disagg：D 侧 KV 容量墙 diff --git a/PAPER_OUTLINE.md b/PAPER_OUTLINE.md index 3e50547..ba90fda 100644 --- a/PAPER_OUTLINE.md +++ b/PAPER_OUTLINE.md @@ -145,7 +145,7 @@ Round-robin 和 load-aware routing（如 LMetric, OSDI'26）最大化 instance **Figure 4: Three baselines, three failure modes** — 拆成三个子图，分别放在 §3.1/§3.2/§3.3： -§3.1 — APC 实测 vs 理论上界 79.6% (lmetric 56.9%, load_only 54.1%, capped 31.6%, sticky 77.2%, unified 79.4%)： +§3.1 — APC 实测 vs 理论上界 79.6% (lmetric 56.9%, load_only 54.1%, sticky 77.2%, unified 79.4%)： ![F4a APC loss](figs/f4a_apc_loss.png) §3.2 — D 侧 KV pool 占用 vs per-request KV footprint，4P+4D 和 6P+2D 在 agentic regime 都穿过 90% 内存墙：