English
Related papers

Related papers: KunServe: Parameter-centric Memory Management for …

200 papers

Serving systems for Large Language Models (LLMs) improve throughput by processing several requests concurrently. However, multiplexing hardware resources between concurrent requests involves non-trivial scheduling decisions. Practical…

Machine Learning · Computer Science 2025-01-29 Ferdi Kossmann , Bruce Fontaine , Daya Khudia , Michael Cafarella , Samuel Madden

Large Language Models (LLMs) have resulted in a surging demand for planet-scale serving systems, where tens of thousands of GPUs continuously serve hundreds of millions of users. Consequently, throughput has emerged as a key metric that…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-05-27 Kan Zhu , Yufei Gao , Yilong Zhao , Liangyu Zhao , Gefei Zuo , Yile Gu , Dedong Xie , Tian Tang , Qinyu Xu , Zihao Ye , Keisuke Kamahori , Chien-Yu Lin , Ziren Wang , Stephanie Wang , Arvind Krishnamurthy , Baris Kasikci

KV cache has traditionally been stored in GPU memory to accelerate the decoding phase of large language model (LLM) inference. However, it is increasingly necessary to move KV caches outside GPU devices, to enable cache reuse across…

Machine Learning · Computer Science 2025-12-08 Yuhan Liu , Yihua Cheng , Jiayi Yao , Yuwei An , Xiaokun Chen , Shaoting Feng , Yuyang Huang , Samuel Shen , Rui Zhang , Kuntai Du , Junchen Jiang

The transformer is the most critical algorithm innovation of the Nature Language Processing (NLP) field in recent years. Unlike the Recurrent Neural Network (RNN) models, Transformers can process on dimensions of sequence lengths in…

Distributed, Parallel, and Cluster Computing · Computer Science 2021-02-23 Jiarui Fang , Yang Yu , Chengduo Zhao , Jie Zhou

When multiple processor cores (CPUs) and a GPU integrated together on the same chip share the off-chip DRAM, requests from the GPU can heavily interfere with requests from the CPUs, leading to low system performance and starvation of cores.…

Hardware Architecture · Computer Science 2018-05-01 Rachata Ausavarungnirun , Gabriel H. Loh , Lavanya Subramanian , Kevin Chang , Onur Mutlu

Large Language Models (LLMs) have achieved remarkable success across a wide range of tasks, but serving them efficiently at scale remains a critical challenge due to their substantial computational and latency demands. While most existing…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-03-04 Yifan Sun , Gholamreza Haffari , Minxian Xu , Rajkumar Buyya , Adel N. Toosi

Large language models (LLMs) have shown remarkable potential in processing long sequences and complex reasoning tasks, yet efficiently serving these models remains challenging due to the quadratic computational complexity of attention in…

Computation and Language · Computer Science 2025-04-22 Shang Yang , Junxian Guo , Haotian Tang , Qinghao Hu , Guangxuan Xiao , Jiaming Tang , Yujun Lin , Zhijian Liu , Yao Lu , Song Han

Pipeline parallelism (PP) has become a standard technique for scaling large language model (LLM) training across multiple devices. However, despite recent progress in reducing memory consumption through activation offloading, existing…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-10-08 Hongpei Li , Han Zhang , Huikang Liu , Dongdong Ge , Yinyu Ye

The growing demand for large language models (LLMs) requires serving systems to handle many concurrent requests with diverse service level objectives (SLOs). This exacerbates head-of-line (HoL) blocking during the compute-intensive prefill…

Distributed, Parallel, and Cluster Computing · Computer Science 2026-02-19 Chia-chi Hsieh , Zan Zong , Xinyang Chen , Jianjiang Li , Jidong Zhai , Lijie Wen

Pipeline parallelism (PP) is widely used for training large language models (LLMs), yet its scalability is often constrained by high activation memory consumption as the number of in-flight microbatches grows with the degree of PP. In this…

Machine Learning · Computer Science 2025-07-01 Xinyi Wan , Penghui Qi , Guangxing Huang , Min Lin , Jialin Li

Supporting long-context LLMs is challenging due to the substantial memory demands of the key-value (KV) cache. Existing offloading systems store the full cache in host memory and selectively fetch critical entries during decoding, but this…

Computation and Language · Computer Science 2026-05-19 Jian Lin , Jiazhi Mi , Zicong Hong , Haodong Wang , Qianli Liu , Haodyue Zhang , Peng Li , Song Guo

Transformers and LLMs have seen rapid adoption in all domains. Their sizes have exploded to hundreds of billions of parameters and keep increasing. Under these circumstances, the training of transformers is slow and often takes in the order…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-06-18 Avinash Maurya , Jie Ye , M. Mustafa Rafique , Franck Cappello , Bogdan Nicolae

Transformer-based large language models (LLMs) have already achieved remarkable results on long-text tasks, but the limited GPU memory (VRAM) resources struggle to accommodate the linearly growing demand for key-value (KV) cache as the…

Computation and Language · Computer Science 2025-03-21 Shibo Jie , Yehui Tang , Kai Han , Zhi-Hong Deng , Jing Han

We study offline scheduling for large language model (LLM) serving under a fixed KV-cache memory budget, where requests have heterogeneous prompt (prefill) and response (decode) lengths. Prompt tokens determine initial KV usage, and each…

Optimization and Control · Mathematics 2026-02-11 Meixuan Wang , Yinyu Ye , Zijie Zhou

KV cache offloading enables long-context LLM inference by storing caches in CPU DRAM, but PCIe bandwidth limitations create severe bottlenecks. In this paper, we develops an analytical framework that derives $\kappa_{\text{crit}}$, the…

Hardware Architecture · Computer Science 2026-01-29 William Meng , Benjamin Lee , Hong Wang

Context retrieval systems for LLM inference face a critical challenge: high retrieval latency creates a fundamental tension between waiting for complete context (poor time-to-first-token) and proceeding without it (reduced quality).…

Databases · Computer Science 2026-05-19 Rajveer Bachkaniwala , Chengqi Luo , Richard So , Divya Mahajan , Kexin Rong

Large language model (LLM) inference serving systems are essential to various LLM-based applications. As demand for LLM services continues to grow, scaling these systems to handle high request rates while meeting latency Service-Level…

Machine Learning · Computer Science 2025-04-11 Shihong Gao , Xin Zhang , Yanyan Shen , Lei Chen

Fine-tuning large language models (LLMs) with low-rank adaptations (LoRAs) has become common practice, often yielding numerous copies of the same LLM differing only in their LoRA updates. This paradigm presents challenges for systems that…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-06-02 Rickard Brüel-Gabrielsson , Jiacheng Zhu , Onkar Bhardwaj , Leshem Choshen , Kristjan Greenewald , Mikhail Yurochkin , Justin Solomon

The widespread adoption of Large Language Models (LLMs) has exponentially increased the demand for efficient serving systems. With growing requests and context lengths, key-value (KV)-related operations, including attention computation and…

Hardware Architecture · Computer Science 2026-02-13 Lian Liu , Shixin Zhao , Yutian Zhou , Yintao He , Mengdi Wang , Yinhe Han , Ying Wang

In modern Commercial Off-The-Shelf (COTS) multicore systems, each core can generate many parallel memory requests at a time. The processing of these parallel requests in the DRAM controller greatly affects the memory interference delay…

Distributed, Parallel, and Cluster Computing · Computer Science 2014-07-29 Heechul Yun