Diffusion Model Serving

Follow this topic Ask about this topic

Momentum

1 paper in the last four weeks, with none the four weeks before. 0.0% of all new papers.

Jul 13Week of Sep 28

Latest papers 10

All topics

Denoising Surface: Modeling and Predicting Inference Cost for Diffusion LLM Serving
Sep 30, 2026Haoyu Zheng, Fangcheng Fu, Binhang Yuan +6Cost-Aware Inference Diffusion Model Serving
Wuhan University · Shanghai Jiao Tong University · The Hong Kong University of Science and Technology +2
CORA-Diff: Confidence-Oriented Residual Acceptance for Efficient Diffusion Language Model Inference
Jul 31, 2026Yifan Wu, Yufeng Zhang, Kenli LiDiffusion Model Serving Diffusion Language Model Inference
College of Computer Science, Hunan University Changsha 410082, China
FlashDiff: Efficient Regional Execution and Scheduling for Diffusion Model Serving
Jul 13, 2026Yaqi Qiao, Ping He, Songrun Xie +4Diffusion Model Serving Diffusion Model Inference Acceleration
University of Illinois Urbana-Champaign · Vanderbilt University · HKUST +2
BlockServe: Block-Grained Continuous Batching for High-Throughput Diffusion LLM Serving
Jul 9, 2026Yuanjie Zhu, Liangwei Yang, Ke Xu +4LLM Inference Scheduling Diffusion Model Serving
University of Illinois Chicago, USA
Sangam: Efficiently Serving Diffusion LLMs with the AR Stack
Jul 5, 2026Nitin Kedia, Saurabh Agarwal, Myungjin Lee +1LLM Inference Scheduling LLM Inference Serving
The University of Texas at Austin Austin, United States · Cisco Research Bellevue, United States
DiLaServe: High SLO Attainment Serving for Diffusion Language Models
Jun 27, 2026Tzu-Tao Chang, Benjamin Yuanyang Hong, Kiet Pham +1LLM Serving Diffusion Model Serving
University of Wisconsin-Madison USA
HERALD: High-Throughput Block Diffusion LLM Serving via CPU-GPU Cooperative KV Cache Retrieval
Jun 19, 2026Omin Kwon, Doyeon Kim, Jongseok Park +3KV-Cache Offloading Diffusion Model Serving
Seoul National University Seoul, Republic of Korea · UC Berkeley Berkeley, CA, USA
GF-DiT: Scheduling Parallelism for Diffusion Transformer Serving
Jun 11, 2026Xinwei Qiang, Yifan Hu, Shixuan Sun +6Diffusion Transformer Diffusion Model Serving
Shanghai Jiao Tong University Shanghai, China · Guizhou University Guiyang, China
FLARE: Diffusion for Hybrid Language Model
Jun 1, 2026Yuchen Zhu, Jing Shi, Chongjian Ge +9Diffusion Model Serving Diffusion Language Model Inference
1Adobe Research · 2Georgia Institute of Technology
SynerDiff: Synergetic Continuous Batching for Fast and Parallel Diffusion Model Inference
May 9, 2026Ziqi Zhou, Peng Yang, Yuxin Liang +2Diffusion Model Serving Diffusion Model Inference Acceleration
School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan, China · The State Grid Hubei Electric Power Research Institute, Wuhan, China · Jincheng Yunxiang Big Data Technology Operation Co., Ltd., Shanxi, China

Sep 30, 2026, cs.LGJ/K move · Enter open · S save

Denoising Surface: Modeling and Predicting Inference Cost for Diffusion LLM Serving

Haoyu Zheng, Fangcheng Fu, Binhang Yuan, Yongqiang Zhang+5

Wuhan University · Shanghai Jiao Tong University · The Hong Kong University of Science and Technology+2

Cost-Aware Inference Diffusion Model Serving Diffusion Language Model Inference LLM Request Scheduling

Open paper AskPDF arXiv

As diffusion large language models (dLLMs) become more capable, they are moving from research settings to real-world \textit{serving}, where request management (such as scheduling and resource allocation) relies on accurate estimation of per-request inference cost. However, common cost proxies fall short for dLLMs: output length ignores that one forward pass can unmask multiple tokens, and denoising-step count ignores the \textit{heterogeneous} per-step costs. We observe that the block-autoregressive generation mechanism induces a two-dimensional execution structure over output blocks and within-block denoising steps, whereas these proxies collapse it into a scalar, discarding information essential for characterizing the cost. Motivated by this insight, we propose the Denoising Workload Surface (DWS), which preserves this two-dimensional block-step structure as a probability surface to weight the heterogeneous per-step costs. We then design a coarse-to-fine training scheme that enables a lightweight prompt-only predictor to accurately predict the complex DWS. This predictor runs efficiently even on a single CPU core, avoiding GPU contention with the serving model. Since DWS decouples request-dependent execution behavior from deployment-specific cost factors, the predictor transfers across hardware configurations without retraining. In \textit{real-world} serving experiments, DWS reduces cost-prediction error by up to

2.50\times

over scalar-based predictors, while the DWS-guided shortest-job-first scheduler reduces end-to-end latency by up to

1.92\times

for online chatbots.

Momentum

Related topics

Latest papers 10