---
cited_by:
  - '[[thoughts/inference/deepseek-v4.1-serving#4-pr-reading-order|dsv41 §4]]'
date: '2026-11-11'
kind: pr
local:
  - file:///Users/aarnphm/workspace/vllm/vllm/distributed/parallel_state.py
  - file:///Users/aarnphm/workspace/vllm/vllm/config/vllm.py
  - file:///Users/aarnphm/workspace/vllm/vllm/envs.py
  - file:///Users/aarnphm/workspace/vllm/.buildkite/test_areas/distributed.yaml
merged: '2026-09-13'
number: 56512
source:
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/models/deepseek_v4_1/nvidia/engram.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/models/deepseek_v4_1/common/engram.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/distributed/parallel_state.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/config/engram.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/config/vllm.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/models/deepseek_v4_1/nvidia/model.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/vllm/envs.py
  - https://github.com/vllm-project/vllm/blob/d2d649e674c75425d2d6975c87eb89fd4d55fff8/.buildkite/test_areas/distributed.yaml
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#56512 [DS V4.1][Engram] Support async prefetch for offloaded engram lookups and engram DP sharding'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/56512
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/56512
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/56512.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Three commits: engram DP sharding (from [#56230](https://github.com/aarnphm/aarnphm.github.io/issues/56230)) plus two engram · full dump: [[thoughts/inference/vllm-prs/56512|#56512]]

