English
Related papers

Related papers: Revisiting Disaggregated Large Language Model Serv…

200 papers

Multi-modal Large Language Models (MLLMs) serving systems commonly employ KV-cache compression to reduce memory footprint. However, existing compression methods introduce significant processing overhead and queuing delays, particularly in…

Multimedia · Computer Science 2025-03-12 Jianian Zhu , Hang Wu , Haojie Wang , Yinghui Li , Biao Hou , Ruixuan Li , Jidong Zhai

Large Language Models (LLMs) use key-value (KV) cache to reduce redundant computation in autoregressive generation. However, the KV cache size increases linearly during generation, leading to excessive memory usage, especially for long…

Computation and Language · Computer Science 2025-03-04 Jian Yuan , Ziwei He , Haoli Bai , Jingwen Leng , Bo Jiang

High-throughput inference serving is essential for applications built on large language models (LLMs). Existing serving frameworks reduce request-level and batch-level bubbles through batching and scheduling, but often overlook bubbles…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-05-25 Fengyao Bai , Hongbin Zhang , Zhitao Chen , Jiangsu Du , Zhiguang Chen , Yutong Lu

The development of large language models (LLMs) has significantly expanded model sizes, resulting in substantial GPU memory requirements during inference. The key and value storage of the attention map in the KV (key-value) cache accounts…

Machine Learning · Computer Science 2024-10-25 Yifei Yang , Zouying Cao , Qiguang Chen , Libo Qin , Dongjie Yang , Hai Zhao , Zhi Chen

Large Language Models (LLMs) demonstrate exceptional capabilities across various tasks, but their deployment is constrained by high computational and memory costs. Model pruning provides an effective means to alleviate these demands.…

Computation and Language · Computer Science 2025-12-16 Hao Zhang , Mengsi Lyu , Zhuo Chen , Xingrun Xing , Yulong Ao , Yonghua Lin

In this study, we introduce adaptive KV cache compression, a plug-and-play method that reduces the memory footprint of generative inference for Large Language Models (LLMs). Different from the conventional KV cache that retains key and…

Computation and Language · Computer Science 2024-10-31 Suyu Ge , Yunan Zhang , Liyuan Liu , Minjia Zhang , Jiawei Han , Jianfeng Gao

Large language model (LLM) serving has transformed from stateless to stateful systems, utilizing techniques like context caching and disaggregated inference. These optimizations extend the lifespan and domain of the KV cache, necessitating…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-12-24 Cunchen Hu , Heyang Huang , Junhao Hu , Jiang Xu , Xusheng Chen , Tao Xie , Chenxi Wang , Sa Wang , Yungang Bao , Ninghui Sun , Yizhou Shan

Large language models (LLMs) are typically served from clusters of GPUs/NPUs that consist of large number of devices. Unfortunately, communication between these devices incurs significant overhead, increasing the inference latency and cost…

Artificial Intelligence · Computer Science 2025-05-27 Ahmet Caner Yüzügüler , Jiawei Zhuang , Lukas Cavigelli

The key-value (KV) cache is a foundational optimization in Transformer-based large language models (LLMs), eliminating redundant recomputation of past token representations during autoregressive generation. However, its memory footprint…

Machine Learning · Computer Science 2026-03-24 Yichun Xu , Navjot K. Khaira , Tejinder Singh

Serving disaggregated large language models (LLMs) over tens of thousands of xPU devices (GPUs or NPUs) with reliable performance faces multiple challenges. 1) Ignoring the diversity (various prefixes and tidal requests), treating all the…

Large language models (LLMs) demonstrate remarkable capabilities but face substantial serving costs due to their high memory demands, with the key-value (KV) cache being a primary bottleneck. State-of-the-art KV cache compression…

Machine Learning · Computer Science 2025-09-03 Yanqi Zhang , Yuwei Hu , Runyuan Zhao , John C. S. Lui , Haibo Chen

How to efficiently serve Large Language Models (LLMs) has become a pressing issue because of their huge computational cost in their autoregressive generation process. To mitigate computational costs, LLMs often employ the KV Cache technique…

Computation and Language · Computer Science 2024-07-23 Zheng Wang , Boxiao Jin , Zhongzhi Yu , Minjia Zhang

Recent advances in Post-Training Quantization (PTQ) techniques have significantly increased demand for serving quantized large language models (LLMs), enabling higher throughput and substantially reduced memory usage with minimal accuracy…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-09-16 Kyungmin Bin , Seungbeom Choi , Jimyoung Son , Jieun Choi , Daseul Bae , Daehyeon Baek , Kihyo Moon , Minsung Jang , Hyojung Lee

Disaggregating the prefill and decoding phases represents an effective new paradigm for generative inference of large language models (LLM), which eliminates prefill-decoding interference and optimizes resource allocation. However, it is…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-02-13 Youhe Jiang , Ran Yan , Binhang Yuan

Recently, sharing key-value (KV) cache across layers has been found effective in efficient inference of large language models (LLMs). To systematically investigate different techniques of cross-layer KV sharing, we propose a unified…

Computation and Language · Computer Science 2025-02-06 You Wu , Haoyi Wu , Kewei Tu

LLM inference must meet strict latency SLOs (e.g., 100 ms P99 time-between-tokens) while maximizing goodput. Yet, real-world variability in prompt and response lengths skews compute-intensive prefill and memory-bound decode phases, making…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-05-23 Chaoyi Ruan , Yinhe Chen , Dongqi Tian , Yandong Shi , Yongji Wu , Jialin Li , Cheng Li

Diffusion-based large language models (Diffusion LLMs) have shown promise for non-autoregressive text generation with parallel decoding capabilities. However, the practical inference speed of open-sourced Diffusion LLMs often lags behind…

Computation and Language · Computer Science 2025-07-04 Chengyue Wu , Hao Zhang , Shuchen Xue , Zhijian Liu , Shizhe Diao , Ligeng Zhu , Ping Luo , Song Han , Enze Xie

Prefill-Decode (PD) disaggregation has become the standard architecture for modern LLM inference engines, which alleviates the interference of two distinctive workloads. With the growing demand for multi-turn interactions in chatbots and…

Networking and Internet Architecture · Computer Science 2026-05-06 Zongze Li , Jingyu Liu , Zhen Xu , Yineng Zhang , Tahseen Rabbani , Ce Zhang

Distributed LLM serving is costly and often underutilizes hardware accelerators due to three key challenges: bubbles in pipeline-parallel deployments caused by the bimodal latency of prompt and token processing, GPU memory overprovisioning,…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-03-05 Foteini Strati , Sara Mcallister , Amar Phanishayee , Jakub Tarnawski , Ana Klimovic

Multimodal large language model (MLLM) inference splits into two phases with opposing hardware demands: vision encoding is compute-bound, while language generation is memory-bandwidth-bound. We show that under standard transformer KV…

Machine Learning · Computer Science 2026-03-16 Donglin Yu