Dovetail: A CPU/GPU Heterogeneous Speculative Decoding for LLM inference

• URL: http://arxiv.org/abs/2412.18934v2
• Date: Tue, 09 Sep 2025 14:27:58 GMT
• Title: Dovetail: A CPU/GPU Heterogeneous Speculative Decoding for LLM inference
• Authors: Libo Zhang, Zhaoning Zhang, Baizhou Xu, Rui Li, Zhiliang Tian, Songzhu Mei, Dongsheng Li,
• Abstract summary: Dovetail is an inference method that leverages the complementary characteristics of heterogeneous devices and the advantages of speculative decoding.<n>Dovetail achieves inference speedups ranging from 1.79x to 10.1x across different devices, while maintaining consistency and stability in the distribution of generated texts.
• Score: 31.901686946969786
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: With the continuous advancement in the performance of large language models (LLMs), their demand for computational resources and memory has significantly increased, which poses major challenges for efficient inference on consumer-grade devices and legacy servers. These devices typically feature relatively weaker GPUs and stronger CPUs. Although techniques such as parameter offloading and partial offloading can alleviate GPU memory pressure to some extent, their effectiveness is limited due to communication latency and suboptimal hardware resource utilization. To address this issue, we propose Dovetail, a lossless inference acceleration method that leverages the complementary characteristics of heterogeneous devices and the advantages of speculative decoding. Dovetail deploys a draft model on the GPU to perform preliminary predictions, while a target model running on the CPU validates these outputs. By reducing the granularity of data transfer, Dovetail significantly minimizes communication overhead. To further improve efficiency, we optimize the draft model specifically for heterogeneous hardware environments by reducing the number of draft tokens to lower parallel verification latency, increasing model depth to enhance predictive capabilities, and introducing a Dynamic Gating Fusion (DGF) mechanism to improve the integration of feature and embedding information. We conduct comprehensive evaluations of Dovetail across various consumer-grade GPUs, covering multiple tasks and mainstream models. Experimental results on 13B models demonstrate that Dovetail achieves inference speedups ranging from 1.79x to 10.1x across different devices, while maintaining consistency and stability in the distribution of generated texts.

Related papers

• Data Driven Optimization of GPU efficiency for Distributed LLM Adapter Serving [2.6336040306318274]
  Large Language Model (LLM) adapters enable low-cost model specialization.<n>LLM adapters introduce complex caching and scheduling challenges in distributed serving systems where hundreds of adapters must be hosted concurrently.<n>This paper presents a data-driven pipeline that computes an adapter placement that serves the workload with the minimum number of GPU.
  arXiv  Detail & Related papers  (2026-02-27T14:22:51Z)
• Scaling LLM Speculative Decoding: Non-Autoregressive Forecasting in Large-Batch Scenarios [76.85739138203014]
  We present SpecFormer, a novel architecture that accelerates unidirectional and attention mechanisms.<n>We demonstrate that SpecFormer achieves lower training demands and reduced computational costs.
  arXiv  Detail & Related papers  (2025-11-25T14:20:08Z)
• ExpertFlow: Adaptive Expert Scheduling and Memory Coordination for Efficient MoE Inference [8.296993547783808]
  ExpertFlow is a runtime system for MoE inference that combines adaptive expert prefetching and cache-aware routing.<n>Our evaluation demonstrates that ExpertFlow reduces model stall time to less than 0.1% of the baseline.
  arXiv  Detail & Related papers  (2025-10-30T17:29:27Z)
• CSGO: Generalized Optimization for Cold Start in Wireless Collaborative Edge LLM Systems [62.24576366776727]
  We propose a latency-aware scheduling framework to minimize total inference latency.<n>We show that the proposed method significantly reduces cold-start latency compared to baseline strategies.
  arXiv  Detail & Related papers  (2025-08-15T07:49:22Z)
• 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)
• The Larger the Merrier? Efficient Large AI Model Inference in Wireless Edge Networks [56.37880529653111]
  The demand for large computation model (LAIM) services is driving a paradigm shift from traditional cloud-based inference to edge-based inference for low-latency, privacy-preserving applications.<n>In this paper, we investigate the LAIM-inference scheme, where a pre-trained LAIM is pruned and partitioned into on-device and on-server sub-models for deployment.
  arXiv  Detail & Related papers  (2025-05-14T08:18:55Z)
• MoE-Lens: Towards the Hardware Limit of High-Throughput MoE LLM Serving Under Resource Constraints [7.287566040274871]
  MoE-Lens is an inference system designed through holistic performance modeling for resource-constrained environments.<n>It captures the system execution mechanisms to identify the key hardware bottlenecks and accurately predict the achievable throughput.<n> evaluated on diverse MoE models and datasets, MoE-Lens outperforms the state-of-the-art solution by 4.6x on average (up to 25.5x)
  arXiv  Detail & Related papers  (2025-04-12T21:26:56Z)
• Task-Oriented Feature Compression for Multimodal Understanding via Device-Edge Co-Inference [54.53508601749513]
  We propose a task-oriented feature compression (TOFC) method for multimodal understanding in a device-edge co-inference framework.<n>To enhance compression efficiency, multiple entropy models are adaptively selected based on the characteristics of the visual features.<n>Results show that TOFC achieves up to 52% reduction in data transmission overhead and 63% reduction in system latency.
  arXiv  Detail & Related papers  (2025-03-17T08:37:22Z)
• 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. 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)
• Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference [4.497936996651617]
  Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. In this paper, through an in-depth GPU-level analysis, we reveal that large-batch inference remains memory-bound, with most GPU compute capabilities underutilized due to DRAM bandwidth saturation as the primary bottleneck.
  arXiv  Detail & Related papers  (2025-03-11T11:21:35Z)
• DuoDecoding: Hardware-aware Heterogeneous Speculative Decoding with Dynamic Multi-Sequence Drafting [59.57151419673759]
  Speculative decoding presents a draft-then-verify framework that reduces generation latency while maintaining output distribution fidelity. We propose DuoDecoding, a novel approach that strategically deploys the draft and target models on the CPU and GPU respectively. Our method incorporates a hardware-aware optimal draft budget to minimize idle times and employs dynamic multi-sequence drafting to enhance draft quality.
  arXiv  Detail & Related papers  (2025-03-02T08:27:48Z)
• PRESERVE: Prefetching Model Weights and KV-Cache in Distributed LLM Serving [2.7309692684728613]
  Large language models (LLMs) are typically served from clusters of GPU/NPUs that consist of large number of devices.<n>Prior work addressed this issue by overlapping communication with compute, but has severe limitations due to the data dependencies between these operations.<n>We propose PRESERVE, a novel framework that prefetches model weights and KV-cache from off-chip memory to the on-chip cache of AI accelerators.
  arXiv  Detail & Related papers  (2025-01-14T15:14:10Z)
• MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs [55.95879347182669]
  MoE architecture is renowned for its ability to increase model capacity without a proportional increase in inference cost. MoE-Lightning introduces a novel CPU-GPU-I/O pipelining schedule, CGOPipe, with paged weights to achieve high resource utilization. MoE-Lightning can achieve up to 10.3x higher throughput than state-of-the-art offloading-enabled LLM inference systems for Mixtral 8x7B on a single T4 GPU (16GB)
  arXiv  Detail & Related papers  (2024-11-18T01:06:12Z)
• Reducing Data Bottlenecks in Distributed, Heterogeneous Neural Networks [5.32129361961937]
  This paper investigates the impact of bottleneck size on the performance of deep learning models in embedded multicore and many-core systems. We apply a hardware-software co-design methodology where data bottlenecks are replaced with extremely narrow layers to reduce the amount of data traffic. Hardware-side evaluation reveals that higher bottleneck ratios lead to substantial reductions in data transfer volume across the layers of the neural network.
  arXiv  Detail & Related papers  (2024-10-12T21:07:55Z)
• LowFormer: Hardware Efficient Design for Convolutional Transformer Backbones [10.435069781620957]
  Research in efficient vision backbones is evolving into models that are a mixture of convolutions and transformer blocks. We analyze common modules and architectural design choices for backbones not in terms of MACs, but rather in actual throughput and latency. We combine both macro and micro design to create a new family of hardware-efficient backbone networks called LowFormer.
  arXiv  Detail & Related papers  (2024-09-05T12:18:32Z)
• Enabling High-Sparsity Foundational Llama Models with Efficient Pretraining and Deployment [56.44025052765861]
  Large language models (LLMs) have revolutionized Natural Language Processing (NLP), but their size creates computational bottlenecks. We introduce a novel approach to create accurate, sparse foundational versions of performant LLMs. We show a total speedup on CPUs for sparse-quantized LLaMA models of up to 8.6x.
  arXiv  Detail & Related papers  (2024-05-06T16:03:32Z)
• FFSplit: Split Feed-Forward Network For Optimizing Accuracy-Efficiency Trade-off in Language Model Inference [57.119047493787185]
  This paper shows how to reduce model size by 43.1% and bring $1.25sim1.56times$ wall clock time speedup on different hardware with negligible accuracy drop. In practice, our method can reduce model size by 43.1% and bring $1.25sim1.56times$ wall clock time speedup on different hardware with negligible accuracy drop.
  arXiv  Detail & Related papers  (2024-01-08T17:29:16Z)
• Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity [12.663030430488922]
  We propose Flash-LLM for enabling low-cost and highly-efficient large generative model inference on high-performance Cores. At SpMM kernel level, Flash-LLM significantly outperforms the state-of-the-art library, i.e., Sputnik and SparTA by an average of 2.9x and 1.5x, respectively.
  arXiv  Detail & Related papers  (2023-09-19T03:20:02Z)
• 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)
• DeepGEMM: Accelerated Ultra Low-Precision Inference on CPU Architectures using Lookup Tables [49.965024476651706]
  DeepGEMM is a lookup table based approach for the execution of ultra low-precision convolutional neural networks on SIMD hardware. Our implementation outperforms corresponding 8-bit integer kernels by up to 1.74x on x86 platforms.
  arXiv  Detail & Related papers  (2023-04-18T15:13:10Z)
• Energy-efficient Task Adaptation for NLP Edge Inference Leveraging Heterogeneous Memory Architectures [68.91874045918112]
  adapter-ALBERT is an efficient model optimization for maximal data reuse across different tasks. We demonstrate the advantage of mapping the model to a heterogeneous on-chip memory architecture by performing simulations on a validated NLP edge accelerator.
  arXiv  Detail & Related papers  (2023-03-25T14:40:59Z)
• Communication-Efficient TeraByte-Scale Model Training Framework for Online Advertising [32.5337643852876]
  Click-Through Rate (CTR) prediction is a crucial component in the online advertising industry. We identify two major challenges in the existing GPU training for massivescale ad models. We propose a hardware-aware training workflow that couples the hardware topology into the algorithm design.
  arXiv  Detail & Related papers  (2022-01-05T18:09:11Z)
• Heterogeneous CPU+GPU Stochastic Gradient Descent Algorithms [1.3249453757295084]
  We study training algorithms for deep learning on heterogeneous CPU+GPU architectures. Our two-fold objective -- maximize convergence rate and resource utilization simultaneously -- makes the problem challenging. We show that the implementation of these algorithms achieves both faster convergence and higher resource utilization than on several real datasets.
  arXiv  Detail & Related papers  (2020-04-19T05:21:20Z)

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.