---
cited_by:
  - '[[thoughts/papers/xie2026mhcmanifoldconstrainedhyperconnections#16-mhc-manifold-constrained-hyper-connections-with-v41s-delayedsingle-pass-pre|dsv41 §1.6]]'
  - '[[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/utils/deep_gemm.py
merged: '2026-09-15'
number: 56962
source:
  - https://github.com/vllm-project/vllm/blob/bb5507741155a698750bc422cda1f7c57f7d114d/vllm/models/deepseek_v41/nvidia/ops/mega_mhc.py
  - https://github.com/vllm-project/vllm/blob/bb5507741155a698750bc422cda1f7c57f7d114d/vllm/models/deepseek_v41/nvidia/model.py
  - https://github.com/vllm-project/vllm/blob/bb5507741155a698750bc422cda1f7c57f7d114d/vllm/utils/deep_gemm.py
  - https://github.com/vllm-project/vllm/blob/bb5507741155a698750bc422cda1f7c57f7d114d/vllm/models/deepseek_v41/nvidia/ops/__init__.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#56962 [Perf][Kernel] Integrate Mega-mHC from DeepGEMM for DeepSeek V4.1  (reopen of #56255)'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/56962
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/56962
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/56962.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Reopen \[vLLM [#56255](https://github.com/aarnphm/aarnphm.github.io/issues/56255)\] (closed unmerged), rebased on the \[[#56876](https://github.com/aarnphm/aarnphm.github.io/issues/56876)\] base (`1b5a7abb58`) with the model path retargeted to `deepseek_v41`. · full dump: [[thoughts/inference/vllm-prs/56962|#56962]]

