REVIEW 11 cited by
Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
abstract
While Diffusion Transformers (DiTs) have achieved breakthroughs in video generation, this long sequence generation task remains constrained by the quadratic complexity of attention mechanisms, resulting in significant inference latency. Through detailed analysis of attention maps in Video Diffusion Transformer (vDiT), we identify three recurring sparsity patterns: diagonal, multi-diagonal, and vertical-stripe structures. And even 3-6\% attention heads can be skipped. Crucially, these patterns exhibit strong layer-depth and head-position correlations but show limited dependence on the input content. Leveraging these findings, we propose Sparse-vDiT, a sparsity acceleration framework for vDiT comprising: 1) Pattern-optimized sparse kernels that replace dense attention with computationally efficient implementations for each identified sparsity pattern. 2) An offline sparse diffusion search algorithm that selects the optimal sparse computation strategy per layer and head via hardware-aware cost modeling. After determining the optimal configuration, we fuse heads within the same layer that share the same attention strategy, enhancing inference efficiency. Integrated into state-of-the-art vDiT models (CogVideoX1.5, HunyuanVideo, and Wan2.1), Sparse-vDiT achieves 2.09$\times$, 2.38$\times$, and 1.67$\times$ theoretical FLOP reduction, and actual inference speedups of 1.76$\times$, 1.85$\times$, and 1.58$\times$, respectively, while maintaining high visual fidelity, with PSNR values reaching 24.13, 27.09, and 22.59. Our work demonstrates that latent structural sparsity in vDiTs can be systematically exploited for long video synthesis.
Forward citations
Cited by 11 Pith papers
-
TwinQuant: Learnable Subspace Decomposition for 4-Bit LLM Quantization
TwinQuant learns quantization-friendly subspaces for 4-bit LLM weights via manifold optimization and a fused kernel, preserving near-FP16 accuracy with up to 1.8x speedup on LLaMA3 and Qwen3 models.
-
HASTE: Training-Free Video Diffusion Acceleration via Head-Wise Adaptive Sparse Attention
HASTE delivers up to 1.93x speedup on Wan2.1 video DiTs via head-wise adaptive sparse attention using temporal mask reuse and error-guided per-head calibration while preserving video quality.
-
Efficient Video Diffusion Models: Advancements and Challenges
A survey that groups efficient video diffusion methods into four paradigms—step distillation, efficient attention, model compression, and cache/trajectory optimization—and outlines open challenges for practical use.
-
PAI-Studio: Cinematic Video Background Replacement with Camera-Aware Motion
PAI-Studio reformulates cinematic background replacement as in-context conditional generation inside a Diffusion Transformer with bidirectional attention, trained on a new 30K film-sourced dataset, and reports better ...
-
Veda: Scalable Video Diffusion via Distilled Sparse Attention
Veda formulates tile selection in video diffusion attention as a reconstruction problem from full attention maps, using statistics-aware and head-aware scoring to enable high sparsity with maintained quality and hardw...
-
Mixture of Distributions Matters: Dynamic Sparse Attention for Efficient Video Diffusion Transformers
MOD-DiT accelerates video DiTs by extrapolating fitted attention-pattern intensities across denoising steps to build dynamic block masks, achieving 1.8–2.3× speedups on CogVideoX, HunyuanVideo, and Wan2.1.
-
Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers
LLSA uses hierarchical top-K selection plus coarse-token enrichment to make diffusion-transformer attention O(N log N), giving about 6x faster training at 256x256 pixel resolution with FID comparable to full attention.
-
Phase-Aligned RoPE for Mixed-Resolution Diffusion Transformer
Expressing all RoPE positions on the query's grid ('one attention, one scale') plus a small boundary content-exchange step restores mixed-resolution diffusion generation that naive position interpolation destroys.
-
SPADE: An Input-Adaptive Sparse Attention Engine for Fast Video Diffusion Models Inference
SPADE combines static, semi-static, and dynamic block-sparse attention with a cheap SICS-based blocking heuristic to speed up video diffusion inference by up to 1.80x end-to-end.
-
ScalingAttention: Discovering Intrinsic Sparse Attention Topology for Video Diffusion Transformers
ScalingAttention discovers a prompt-agnostic intrinsic sparse attention topology in video DiTs via weight-encoded extraction (WEST) and fidelity-aware tuning (FAST), delivering up to 1.9x speedup.
-
Ride the Wave: Precision-Allocated Sparse Attention for Smooth Video Generation
PASA uses curvature-aware dynamic budgeting, grouped approximations, and stochastic attention routing to accelerate video diffusion transformers while eliminating temporal flickering from sparse patterns.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.