Preprint

SPADE: An Input-Adaptive Sparse Attention Engine for Fast Video Diffusion Models Inference

Shang-Hao Liu Renze Chen Size Zheng Yuanqiang Liu Yun Liang Hailong Yang

Aug 2026 · 1 citation · ⚡ 1 influential · 33 references

Computer Science

TL;DR

SPADE is presented, a training-free sparse-attention engine of three parts: a specification defining 3D blocking candidates and formalizing dynamic masks via Summarizer/Estimator expressions, and an executor with low-overhead index search, flash block-sparse attention, and kernel grouping.

Abstract

Video diffusion transformers (vDiTs) generate high quality but pay quadratic self-attention cost, making inference prohibitive at video-token scales. The challenge is input-adaptive sparsity: selecting critical Q/K/V tokens with negligible overhead and executing them for end-to-end gains. We present SPADE, a training-free sparse-attention engine of three parts: (i) vDiT-SSR, a specification defining 3D blocking candidates and formalizing dynamic masks via Summarizer/Estimator expressions; (ii) runtime scheme generation using SICS and a head-wise policy; and (iii) an executor with low-overhead index search, flash block-sparse attention, and kernel grouping. Across Hunyuan-Video and Wan 2.1/2.2 for text-to-video and image-to-video generation, SPADE raises sparsity and speed while preserving quality, accelerating attention by 2.26x-3.40x and end-to-end inference by 1.49x-1.80x. Our code is open-sourced at https://github.com/6somehow/DAC-SPADE.

DOI arXiv Semantic Scholar

View source

Similar papers

Jul 2026

Sol-Attn: Accelerating Video Generation Inference via On-the-Fly Attention Sparsification

This paper introduces training-free Sol-Attn (Sparsifying online attention), which unifies dynamic routing, sparse computation, and approximation correction in a single online-softmax pass, achieving a better accuracy-efficiency trade-off in sparse attention.

Haopeng Li, Yitong Li, Junsong Chen et al. · arXiv.org · 2 citations

Preprint Jul 2026

MobileWan: Closing the Quality Gap for Mobile Video Diffusion

MobileWan becomes the first 5B-scale video diffusion model deployable on a commercial mobile device and proposes a learnable attention head pruning method based on binary per-head gates optimized end-to-end using a noise-biased sparsity objective and distillation-based finetuning.

Mohsen Ghafoorian, Denis Korzhenkov, Adil Karjauv et al. · 1 citation

Preprint Aug 2026

Token Radius Attention for Efficient Video Generation

This work proposes Token Radius Attention (TRA), a training-free framework that maps query entropy to an analytic token budget and converts it into a temporally decayed radius without explicit key ranking and achieves 1.05x speedup with competitive generation quality.

Jia-Yu Chen, Zhi-Kun Jiang, Mao-Liang Li et al. · 0 citations

Preprint Jul 2026

HyperVAttention: Efficient Sparse Attention with Spatio-Temporal Clustering for Video Diffusion

Experiments show that HVA establishes a new Pareto frontier for training-free sparse attention in video diffusion, reducing end-to-end latency by up to $2.13\times while improving fidelity over existing training-free sparse attention baselines.

Dongyeun Lee, A. Zandieh, V. Mirrokni et al. · 1 citation

Jul 2026

LaCache: Exact Caching and Precision-Adaptive Inference for Diffusion Large Language Models

LaCache is proposed, a training-free acceleration framework that alleviates operator-level redundancy through lossless caching and mixed precision, and inegrates a per-group FP8 quantization strategy for FFN layers, tailored to step-dependent activation distributions across the diffusion process.

Xingru Chen, Zelang Liang, Yongjia Ma et al. · arXiv.org · 0 citations

Conference Jul 2026

AST-ToMe: adaptive step-aware thresholding for efficient temporal-spatial feature reuse in diffusion models

The exceptional generative capabilities of modern diffusion models are fundamentally bottlenecked by the quadratic computational complexity of their attention mechanisms. While recent feature caching strategies attempt to accelerate inference by skipping layers at static intervals, they fail to account for the non-line...

Shu-Zhi Zheng · International Conference on... · 0 citations