Robust Parallel Diffusion Sampling via Dynamic Jacobian Bandwidth

Recent years have witnessed significant progress in developing effective diffusion models. Parallel sampling is a promising recent approach that reformulates the sequential denoising process as solving a system of nonlinear equations, and it can be combined with other acceleration techniques. However, current progress is limited by the trade-off between high fidelity and computational efficiency. This paper addresses the challenge of scaling to high-dimensional, multi-modal generation. Specifically, we present ROPA (Robust Parallel Diffusion Sampling), which takes into account the properties of the denoising process and solves the linear system using adaptive local sparsity to achieve stable parallel sampling. Extensive experiments demonstrate ROPA’s effectiveness: it significantly accelerates sampling across diverse image and video diffusion models, achieving up to $2.9\times$ speedup with eight core, an improvement of 20.8\% over baselines without sacrificing sample quality. ROPA enables parallel sampling methods to provide a solid foundation for real-time, high-fidelity diffusion generation.

Diffusion models can generate high-quality images and videos, but they are often slow because they usually create samples through many steps that must be run one after another. This makes them difficult to use in real-time or large-scale applications. Our paper introduces ROPA, a new method that makes this generation process faster by allowing multiple steps to be solved in parallel while still keeping the final image or video quality high. The key idea is to adapt how much information is shared across generation steps: ROPA uses more careful computation in difficult parts of the process and simpler computation where the process is already stable. This helps avoid errors that can appear when generation is sped up too aggressively. Across several large image and video generation models, ROPA substantially reduces generation time while preserving visual quality, making fast and reliable diffusion-based generation more practical.