大模型
arxiv.org ·
PipeLive：面向动态 LLM 服务的高效实时原地流水线并行重配置
该论文针对现有流水线并行（PP）配置静态、难以适应 serverless 与异构 GPU 等动态场景的问题，提出 PipeLive，实现不中断推理的原地在线 PP 重配置。其核心是重设计的 KV cache 布局与协同扩展的 PageAttention，形成统一的在线 KV 重缩放机制，并借鉴虚拟机在线迁移思想引入增量 KV 补丁，在源与目标配置间同步 KV 状态并确定安全切换点。作者报告：相比关闭 KV resizing，TTFT 降低 2.5 倍且无 KV cache 溢出；相比无 KV 补丁的变体，重配置开销从秒级降至 10ms 以内，TTFT 与 TPOT 分别最高改善 54.7% 和 14.7%。
AI推荐理由 arXiv 论文（cs.DC/cs.LG），摘要给出明确机制与量化结果：重配置开销从秒级降至 10ms 以内，TTFT 最高改善 54.7%、TPOT 14.7%，相比关闭 KV resizing 时 TTFT 降低 2.5 倍。新信息在于面向动态 LLM 服务的原地在线流水线并行重配置，结合重设计的 KV cache 布局与 PageAttention 扩展、借鉴虚拟机在线迁移的增量 KV 补丁机制。局限：结果均为作者自报，尚无第三方复现或生产部署证据，且未说明测试模型规模与硬件配置。
78
正文
Computer Science > Distributed, Parallel, and Cluster Computing
arXiv:2604.12171 (cs)
[Submitted on 14 Apr 2026 (v1), last revised 23 Sep 2026 (this version, v2)]
Title:PipeLive: Efficient Live In-place Pipeline Parallelism Reconfiguration for Dynamic LLM Serving
Authors:Xu Bai, Muhammed Tawfiqul Islam, Chen Wang, Adel N. Toosi
View a PDF of the paper titled PipeLive: Efficient Live In-place Pipeline Parallelism Reconfiguration for Dynamic LLM Serving, by Xu Bai and 3 other authors
View PDF
HTML (experimental)
Abstract:Pipeline parallelism (PP) is widely used to partition layers of large language models (LLMs) across GPUs, enabling scalable inference for large models. However, existing systems rely on static PP configurations that fail to adapt to dynamic settings, such as serverless platforms and heterogeneous GPU environments. Reconfiguring PP by stopping and redeploying service incurs prohibitive downtime, so reconfiguration must instead proceed live and in place, without interrupting inference. However, live in-place PP reconfiguration is fundamentally challenging. GPUs are already saturated with model weights and KV cache, leaving little room for new layer placements and necessitating KV cache resizing, at odds with systems like vLLM that preallocate for throughput. Moreover, maintaining KV consistency during execution is difficult: stop-and-copy introduces large pauses, while background synchronization risks inconsistency as states evolve. We present PipeLive, which enables live in-place PP reconfiguration with minimal disruption. PipeLive introduces a redesigned KV cache layout together with a co-designed extension to PageAttention, forming a unified mechanism for live KV resizing. It further adopts an incremental KV patching mechanism, inspired by live virtual machine migration, to synchronize KV states between source and target configurations and identify a safe switch point. PipeLive achieves a 2.5X reduction in time-to-first-token (TTFT) without KV cache overflow compared to disabling KV resizing. Furthermore, compared to a variant without KV patching, it reduces reconfiguration overhead from seconds to under 10ms, and improves TTFT and time-per-output-token (TPOT) by up to 54.7% and 14.7%, respectively.
Subjects:
Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
Cite as:
arXiv:2604.12171 [cs.DC]
(or
arXiv:2604.12171v2 [cs.DC] for this version)
https://doi.org/10.48550/arXiv.2604.12171
Focus to learn more
arXiv-issued DOI via DataCite
Submission history
From: Xu Bai [view email]
[v1]
Tue, 14 Apr 2026 00:56:32 UTC (494 KB)
[v2]
Wed, 23 Sep 2026 10:43:25 UTC (453 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled PipeLive: Efficient Live In-place Pipeline Parallelism Reconfiguration for Dynamic LLM Serving, by Xu Bai and 3 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.DC
new
|
recent
| 2026-04
Change to browse by:
cs
cs.LG
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
export BibTeX citation
Loading...
BibTeX formatted citation
×
loading...
Data provided by:
Bookmark
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)