DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models

TL;DR

DistriFusion introduces a parallel inference method for high-resolution diffusion models that leverages patch-based model splitting and reused features to significantly accelerate image synthesis without quality loss.

Contribution

It proposes displaced patch parallelism, enabling efficient distributed inference of diffusion models through asynchronous communication and feature reuse.

Findings

Achieves up to 6.1× speedup on eight GPUs

No quality degradation on Stable Diffusion XL

Supports asynchronous communication for efficient parallelism

Abstract

Diffusion models have achieved great success in synthesizing high-quality images. However, generating high-resolution images with diffusion models is still challenging due to the enormous computational costs, resulting in a prohibitive latency for interactive applications. In this paper, we propose DistriFusion to tackle this problem by leveraging parallelism across multiple GPUs. Our method splits the model input into multiple patches and assigns each patch to a GPU. However, naively implementing such an algorithm breaks the interaction between patches and loses fidelity, while incorporating such an interaction will incur tremendous communication overhead. To overcome this dilemma, we observe the high similarity between the input from adjacent diffusion steps and propose displaced patch parallelism, which takes advantage of the sequential nature of the diffusion process by reusing the pre-computed feature maps from the previous timestep to provide context for the current step. Therefore, our method supports asynchronous communication, which can be pipelined by computation. Extensive experiments show that our method can be applied to recent Stable Diffusion XL with no quality degradation and achieve up to a 6.1$\times$ speedup on eight NVIDIA A100s compared to one. Our code is publicly available at https://github.com/mit-han-lab/distrifuser.