---
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:
  - file:///Users/aarnphm/workspace/vllm/vllm/model_executor/layers/attention/mla_attention.py
  - file:///Users/aarnphm/workspace/vllm/vllm/v1/attention/ops/common.py
merged: '2026-08-10'
number: 50484
source:
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/vllm/v1/attention/ops/dcp_utils.py
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/csrc/libtorch_stable/attention/dcp_utils/dcp_direct_a2a_lse_reduce.cu
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/csrc/libtorch_stable/attention/dcp_utils/dcp_direct_kv_gather.cu
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/csrc/libtorch_stable/attention/dcp_utils/dcp_direct_q_gather.cu
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/vllm/model_executor/layers/attention/mla_attention.py
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/csrc/libtorch_stable/attention/dcp_utils/dcp_direct_common.cuh
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/vllm/models/kimi_k3/nvidia/mla.py
  - https://github.com/vllm-project/vllm/blob/63ac04a61e6272b746d82e180e7f6276b3ee1ede/vllm/v1/attention/ops/common.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#50484 [Kimi-K3] DCP support'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/50484
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/50484
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/50484.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
direct symmetric-memory DCP A2A output/LSE reduction, including empty-KV-shard masking · full dump: [[thoughts/inference/vllm-prs/50484|#50484]]

