PipeFusion: Patch-level Pipeline Parallelism for Diffusion Transformers Inference

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fang, Jiarui, Pan, Jinzhe, Li, Aoyu, Sun, Xibo, Wang, Jiannan
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914526604034048
author Fang, Jiarui
Pan, Jinzhe
Li, Aoyu
Sun, Xibo
Wang, Jiannan
author_facet Fang, Jiarui
Pan, Jinzhe
Li, Aoyu
Sun, Xibo
Wang, Jiannan
contents This paper presents PipeFusion, an innovative parallel methodology to tackle the high latency issues associated with generating high-resolution images using diffusion transformers (DiTs) models. PipeFusion partitions images into patches and the model layers across multiple GPUs. It employs a patch-level pipeline parallel strategy to orchestrate communication and computation efficiently. By capitalizing on the high similarity between inputs from successive diffusion steps, PipeFusion reuses one-step stale feature maps to provide context for the current pipeline step. This approach notably reduces communication costs compared to existing DiTs inference parallelism, including tensor parallel, sequence parallel and DistriFusion. PipeFusion enhances memory efficiency through parameter distribution across devices, ideal for large DiTs like Flux.1. Experimental results demonstrate that PipeFusion achieves state-of-the-art performance on 8$\times$L40 PCIe GPUs for Pixart, Stable-Diffusion 3, and Flux.1 models. Our source code is available at https://github.com/xdit-project/xDiT.
format Preprint
id arxiv_https___arxiv_org_abs_2405_14430
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle PipeFusion: Patch-level Pipeline Parallelism for Diffusion Transformers Inference
Fang, Jiarui
Pan, Jinzhe
Li, Aoyu
Sun, Xibo
Wang, Jiannan
Computer Vision and Pattern Recognition
Artificial Intelligence
Performance
This paper presents PipeFusion, an innovative parallel methodology to tackle the high latency issues associated with generating high-resolution images using diffusion transformers (DiTs) models. PipeFusion partitions images into patches and the model layers across multiple GPUs. It employs a patch-level pipeline parallel strategy to orchestrate communication and computation efficiently. By capitalizing on the high similarity between inputs from successive diffusion steps, PipeFusion reuses one-step stale feature maps to provide context for the current pipeline step. This approach notably reduces communication costs compared to existing DiTs inference parallelism, including tensor parallel, sequence parallel and DistriFusion. PipeFusion enhances memory efficiency through parameter distribution across devices, ideal for large DiTs like Flux.1. Experimental results demonstrate that PipeFusion achieves state-of-the-art performance on 8$\times$L40 PCIe GPUs for Pixart, Stable-Diffusion 3, and Flux.1 models. Our source code is available at https://github.com/xdit-project/xDiT.
title PipeFusion: Patch-level Pipeline Parallelism for Diffusion Transformers Inference
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Performance
url https://arxiv.org/abs/2405.14430