Nexus:Proactive Intra-GPU Disaggregation of Prefill and Decode in LLM Serving

• URL: http://arxiv.org/abs/2507.06608v5
• Date: Thu, 07 Aug 2025 12:26:15 GMT
• Title: Nexus:Proactive Intra-GPU Disaggregation of Prefill and Decode in LLM Serving
• Authors: Xiaoxiang Shi, Colin Cai, Junjia Du, Zhihao Jia,
• Abstract summary: Engine-level prefill-decode (PD) disaggregation avoids interference but incurs higher hardware and coordination overhead.<n>PD achieves up to 2.2x higher throughput, 20x lower TTFT, and 2.5x lower TBT than vLLM; outperforms SG by up to 2x; and matches or exceeds disaggregated vLLM.
• Score: 4.309392302169281
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Monolithic serving with chunked prefill improves GPU utilization by batching prefill and decode together, but suffers from fine-grained phase interference. Engine-level prefill-decode (PD) disaggregation avoids interference but incurs higher hardware and coordination overhead. Prior intra-GPU disaggregation approaches multiplex prefill and decode within a single GPU, using SLO-based tuning guided by heuristics from offline profiling or reactive feedback loops. However, these methods respond reactively to performance issues rather than anticipating them, limiting adaptability under dynamic workloads. We ask: can we achieve proactive intra-GPU disaggregation that adapts effectively to dynamic workloads? The key challenge lies in managing the conflicting resource demands of prefill and decode under varying conditions. We first show that GPU resources exhibit diminishing returns -- beyond a saturation point, more allocation yields minimal latency benefit. Second, we observe that memory bandwidth contention becomes a critical bottleneck. These insights motivate a design that dynamically partitions GPU resources across prefill and decode phases, while jointly considering compute capacity, memory footprint, and bandwidth contention. Evaluated on diverse LLMs and workloads, our system Nexus achieves up to 2.2x higher throughput, 20x lower TTFT, and 2.5x lower TBT than vLLM; outperforms SGLang by up to 2x; and matches or exceeds disaggregated vLLM.

Related papers

• Hardware-Efficient Attention for Fast Decoding [13.958883001629644]
  Grouped Latent Attention (GLA) is a parallel-friendly latent attention paired with low-level optimizations for fast decoding.<n>Our optimized GLA kernel is up to 2$times$ faster than FlashMLA, for example, in a speculative decoding setting.
  arXiv  Detail & Related papers  (2025-05-27T17:54:07Z)
• Minute-Long Videos with Dual Parallelisms [57.22737565366549]
  Diffusion Transformer (DiT)-based video diffusion models generate high-quality videos at scale but incur prohibitive processing latency and memory costs for long videos.<n>We propose a novel distributed inference strategy, termed DualParal.<n>Instead of generating an entire video on a single GPU, we parallelize both temporal frames and model layers across GPUs.
  arXiv  Detail & Related papers  (2025-05-27T11:55:22Z)
• FastCar: Cache Attentive Replay for Fast Auto-Regressive Video Generation on the Edge [60.000984252907195]
  Auto-regressive (AR) models have recently shown promise in visual generation tasks due to their superior sampling efficiency.<n>Video generation requires a substantially larger number of tokens to produce coherent temporal frames, resulting in significant overhead during the decoding phase.<n>We propose the textbfFastCar framework to accelerate the decode phase for the AR video generation by exploring the temporal redundancy.
  arXiv  Detail & Related papers  (2025-05-17T05:00:39Z)
• SpecOffload: Unlocking Latent GPU Capacity for LLM Inference on Resource-Constrained Devices [16.407669822378487]
  SpecOffload embeds speculative decoding into offloading.<n>Compared to the best baseline, SpecOffload improves GPU core utilization by 4.49x and boosts inference throughput by 2.54x.
  arXiv  Detail & Related papers  (2025-05-15T13:10:31Z)
• ZO2: Scalable Zeroth-Order Fine-Tuning for Extremely Large Language Models with Limited GPU Memory [29.245719403159615]
  We propose a novel framework, ZO2, for efficient zeroth-order fine-tuning of LLMs with only limited GPU memory.<n>Our framework supports an innovative low-bit precision approach in AMP mode to streamline data exchanges between the CPU and GPU.
  arXiv  Detail & Related papers  (2025-03-16T21:58:29Z)
• POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference [9.164093249308419]
  We present POD-Attention - the first GPU kernel that efficiently computes attention for hybrid batches.<n> POD-Attention aims to maximize the utilization of both compute and memory bandwidth by carefully allocating the GPU's resources.
  arXiv  Detail & Related papers  (2024-10-23T17:06:56Z)
• MARLIN: Mixed-Precision Auto-Regressive Parallel Inference on Large Language Models [58.3342517278868]
  This paper describes the design of Mixed-precision AutoRegressive LINear kernels. It shows that batchsizes up to 16-32 can be supported with close to maximum ($4times$) quantization speedup. MarLIN accomplishes this via a combination of techniques, such as asynchronous memory access, complex task scheduling and pipelining.
  arXiv  Detail & Related papers  (2024-08-21T16:10:41Z)
• vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving [53.972175896814505]
  Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests.
  arXiv  Detail & Related papers  (2024-07-22T14:37:58Z)
• FusionAI: Decentralized Training and Deploying LLMs with Massive Consumer-Level GPUs [57.12856172329322]
  We envision a decentralized system unlocking the potential vast untapped consumer-level GPU. This system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity.
  arXiv  Detail & Related papers  (2023-09-03T13:27:56Z)
• SqueezeLLM: Dense-and-Sparse Quantization [80.32162537942138]
  Main bottleneck for generative inference with LLMs is memory bandwidth, rather than compute, for single batch inference. We introduce SqueezeLLM, a post-training quantization framework that enables lossless compression to ultra-low precisions of up to 3-bit. Our framework incorporates two novel ideas: (i) sensitivity-based non-uniform quantization, which searches for the optimal bit precision assignment based on second-order information; and (ii) the Dense-and-Sparse decomposition that stores outliers and sensitive weight values in an efficient sparse format.
  arXiv  Detail & Related papers  (2023-06-13T08:57:54Z)
• FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU [89.2451963569343]
  FlexGen is a generation engine for running large language model (LLM) inference on a single commodity GPU. When running OPT-175B on a single 16GB GPU, FlexGen achieves significantly higher throughput compared to state-of-the-art offloading systems. On the HELM benchmark, FlexGen can benchmark a 30B model with a 16GB GPU on 7 representative sub-scenarios in 21 hours.
  arXiv  Detail & Related papers  (2023-03-13T05:19:28Z)
• Heterogeneous Acceleration Pipeline for Recommendation System Training [1.8457649813040096]
  Recommendation models rely on deep learning networks and large embedding tables. These models are typically trained using hybrid-GPU or GPU-only configurations. This paper introduces Hotline, a heterogeneous CPU acceleration pipeline.
  arXiv  Detail & Related papers  (2022-04-11T23:10:41Z)
• PARIS and ELSA: An Elastic Scheduling Algorithm for Reconfigurable Multi-GPU Inference Servers [0.9854614058492648]
  NVIDIA's Ampere GPU architecture provides features to "reconfigure" one large, monolithic GPU into multiple smaller "GPU partitions" In this paper, we study this emerging GPU architecture with reconfigurability to develop a high-performance multi-GPU ML inference server.
  arXiv  Detail & Related papers  (2022-02-27T23:30:55Z)
• PLSSVM: A (multi-)GPGPU-accelerated Least Squares Support Vector Machine [68.8204255655161]
  Support Vector Machines (SVMs) are widely used in machine learning. However, even modern and optimized implementations do not scale well for large non-trivial dense data sets on cutting-edge hardware. PLSSVM can be used as a drop-in replacement for an LVM.
  arXiv  Detail & Related papers  (2022-02-25T13:24:23Z)
• Adaptive Elastic Training for Sparse Deep Learning on Heterogeneous Multi-GPU Servers [65.60007071024629]
  We show that Adaptive SGD outperforms four state-of-the-art solutions in time-to-accuracy. We show experimentally that Adaptive SGD outperforms four state-of-the-art solutions in time-to-accuracy.
  arXiv  Detail & Related papers  (2021-10-13T20:58:15Z)
• Nimble: Lightweight and Parallel GPU Task Scheduling for Deep Learning [7.43260596107574]
  We propose Nimble, a deep learning (DL) execution engine that runs tasks in parallel with minimal scheduling overhead. Nable automatically parallelizes the execution of GPU tasks by exploiting multiple GPU streams in a single GPU. evaluation on a variety of neural networks shows that compared to PyTorch, Nimble speeds up inference and training by up to 22.34$times$ and 3.61$times$, respectively.
  arXiv  Detail & Related papers  (2020-12-04T17:25:46Z)
• GPU-Accelerated Primal Learning for Extremely Fast Large-Scale Classification [10.66048003460524]
  One of the most efficient methods to solve L2-regularized primal problems, such as logistic regression and linear support vector machine (SVM) classification, is the widely used trust region Newton algorithm, TRON. We show that using judicious GPU-optimization principles, TRON training time for different losses and feature representations may be drastically reduced.
  arXiv  Detail & Related papers  (2020-08-08T03:40:27Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.