Efficient Multi-round LLM Inference over Disaggregated Serving
Shanghai Jiao Tong University（上海交通大学）；Southeast University（东南大学）；University of Cambridge（剑桥大学）；Peking University（北京大学）；OceanBase, Ant Group（蚂蚁集团 OceanBase）；Computational Social Science, Peking University (Qingdao)（北京大学（青岛）计算社会科学）
Wenhao He, Youhe Jiang, Penghao Zhao, Quanqing Xu, Eiko Yoneki, Bin Cui, Fangcheng Fu
With the rapid evolution of Large Language Models (LLMs), multi-round workflows, such as autonomous agents and iterative retrieval, have become increasingly prevalent. However, this raises hurdles for serving LLMs under prefill-decode (PD) disaggregation, a widely adopted paradigm that separates the compute-bound prefill phase and memory-bound decode phase onto individual resources. Specifically, existing systems overlook the interleaved prefill-decode workload pattern in multi-round inference, leading to sub-optimal handling of the incremental prefill workloads and model deployment for the two phases.
In this work, we present AMPD, a brand new disaggregated serving framework for multi-round LLM inference. The core of AMPD is to coordinate the prefill workloads based on real-time workloads by adaptively determining where to carry out these workloads and how they are scheduled, in order to maximize service level objective (SLO) attainment. In addition, we tailor a planning algorithm for our scenario, facilitating the deduction of optimal resource allocation and parallel strategies for the two phases. Empirical results demonstrate that AMPD substantially improves SLO attainment compared to state-of-the-art baselines.
Tropical: Enhancing SLO Attainment in Disaggregated LLM Serving via SLO-Aware Multiplexing
Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）；Fudan University（复旦大学）；Peking University（北京大学）；The Chinese University of Hong Kong（香港中文大学）；Shanghai Jiao Tong University（上海交通大学）；Sensetime Research（商汤科技研究院）
InfiniLoRA: Disaggregated Multi-LoRA Serving for Large Language Models
Shanghai Jiao Tong University（上海交通大学）；Bytedance（字节跳动）；Singapore Management University（新加坡管理大学）；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；National University of Singapore（新加坡国立大学）
DisagFusion: Asynchronous Pipeline Parallelism and Elastic Scheduling for Disaggregated Diffusion Serving
Renmin University of China（中国人民大学）；SenseTime（商汤科技）；Peking University（北京大学）；Hong Kong University of Science and Technology（香港科技大学）；Beihang University（北京航空航天大学）
vLLM-Omni: Fully Disaggregated Serving for Any-to-Any Multimodal Models
AI Framework and Data Technology Lab, Huawei（华为人工智能框架与数据技术实验室）；The Chinese University of Hong Kong（香港中文大学）；Institute of Software, Chinese Academy of Sciences（中国科学院软件研究所）；Sun Yat-sen University（中山大学）
BanaServe: Unified KV Cache and Dynamic Module Migration for Balancing Disaggregated LLM Serving in AI Infrastructure
Southern University of Science and Technology（南方科技大学）；Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究院）；University of Chinese Academy of Sciences（中国科学院大学）；State Key Lab of IOTSC, Faculty of Science and Technology, University of Macau（澳门科技大学物联网与感知系统国家重点实验室、理学院）；AIOS Team, Alibaba Group Inc（阿里巴巴集团 AIOS 团队）
KVServe: Service-Aware KV Cache Compression for Communication-Efficient Disaggregated LLM Serving
University of Chinese Academy of Sciences（中国科学院大学）；Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；Shanghai Jiao Tong University（上海交通大学）
Stage-Aware Communication Scheduling for Disaggregated LLM Serving
iSING Lab, Hong Kong University of Science and Technology（香港科技大学 iSING 实验室）；Shanghai Jiao Tong University（上海交通大学）；University of Science and Technology of China（中国科学技术大学）
