---
cited_by:
  - '[[thoughts/inference/kimi-k3-state#34-pd-dcp-kv-offload-59229-test-matrix|kimik3 §3.4]]'
  - '[[thoughts/inference/kimi-k3-prs#4-pr-reading-order|kimik3 §4]]'
date: '2026-11-11'
kind: pr
local: []
merged: '2026-08-27'
number: 53396
source:
  - https://github.com/vllm-project/vllm/blob/aa640684cc95f1e4b14fdf38298ea8577a70de56/csrc/libtorch_stable/kimi_k3/fused_kda_decode_kernel.cu
  - https://github.com/vllm-project/vllm/blob/aa640684cc95f1e4b14fdf38298ea8577a70de56/csrc/libtorch_stable/kimi_k3/fused_kda_decode_kernel_rocm.cu
  - https://github.com/vllm-project/vllm/blob/aa640684cc95f1e4b14fdf38298ea8577a70de56/vllm/models/kimi_k3/nvidia/kda.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#53396 [Kimi K3][Kernel] Support DS conv-state layout in fused KDA decode kernel'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/53396
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/53396
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/53396.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
The fused KDA decode kernel (`ops.fused_kda_decode`) previously required the SD conv-state cache layout, so any deployment that pins `VLLM_SSM_CONV_STATE_LAYOUT=DS` — required for NIXL point-to-point state transfer in P/D-disaggregated… · full dump: [[thoughts/inference/vllm-prs/53396|#53396]]

