DDiT: Dynamic Resource Allocation for Diffusion Transformer Model Serving

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Heyang, Hu, Cunchen, Zhu, Jiaqi, Gao, Ziyuan, Xu, Liangliang, Shan, Yizhou, Bao, Yungang, Ninghui, Sun, Zhang, Tianwei, Wang, Sa
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911007491751936
author Huang, Heyang
Hu, Cunchen
Zhu, Jiaqi
Gao, Ziyuan
Xu, Liangliang
Shan, Yizhou
Bao, Yungang
Ninghui, Sun
Zhang, Tianwei
Wang, Sa
author_facet Huang, Heyang
Hu, Cunchen
Zhu, Jiaqi
Gao, Ziyuan
Xu, Liangliang
Shan, Yizhou
Bao, Yungang
Ninghui, Sun
Zhang, Tianwei
Wang, Sa
contents The Text-to-Video (T2V) model aims to generate dynamic and expressive videos from textual prompts. The generation pipeline typically involves multiple modules, such as language encoder, Diffusion Transformer (DiT), and Variational Autoencoders (VAE). Existing serving systems often rely on monolithic model deployment, while overlooking the distinct characteristics of each module, leading to inefficient GPU utilization. In addition, DiT exhibits varying performance gains across different resolutions and degrees of parallelism, and significant optimization potential remains unexplored. To address these problems, we present DDiT, a flexible system that integrates both inter-phase and intra-phase optimizations. DDiT focuses on two key metrics: optimal degree of parallelism, which prevents excessive parallelism for specific resolutions, and starvation time, which quantifies the sacrifice of each request. To this end, DDiT introduces a decoupled control mechanism to minimize the computational inefficiency caused by imbalances in the degree of parallelism between the DiT and VAE phases. It also designs a greedy resource allocation algorithm with a novel scheduling mechanism that operates at the single-step granularity, enabling dynamic and timely resource scaling. Our evaluation on the T5 encoder, OpenSora SDDiT, and OpenSora VAE models across diverse datasets reveals that DDiT significantly outperforms state-of-the-art baselines by up to 1.44x in p99 latency and 1.43x in average latency.
format Preprint
id arxiv_https___arxiv_org_abs_2506_13497
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DDiT: Dynamic Resource Allocation for Diffusion Transformer Model Serving
Huang, Heyang
Hu, Cunchen
Zhu, Jiaqi
Gao, Ziyuan
Xu, Liangliang
Shan, Yizhou
Bao, Yungang
Ninghui, Sun
Zhang, Tianwei
Wang, Sa
Distributed, Parallel, and Cluster Computing
The Text-to-Video (T2V) model aims to generate dynamic and expressive videos from textual prompts. The generation pipeline typically involves multiple modules, such as language encoder, Diffusion Transformer (DiT), and Variational Autoencoders (VAE). Existing serving systems often rely on monolithic model deployment, while overlooking the distinct characteristics of each module, leading to inefficient GPU utilization. In addition, DiT exhibits varying performance gains across different resolutions and degrees of parallelism, and significant optimization potential remains unexplored. To address these problems, we present DDiT, a flexible system that integrates both inter-phase and intra-phase optimizations. DDiT focuses on two key metrics: optimal degree of parallelism, which prevents excessive parallelism for specific resolutions, and starvation time, which quantifies the sacrifice of each request. To this end, DDiT introduces a decoupled control mechanism to minimize the computational inefficiency caused by imbalances in the degree of parallelism between the DiT and VAE phases. It also designs a greedy resource allocation algorithm with a novel scheduling mechanism that operates at the single-step granularity, enabling dynamic and timely resource scaling. Our evaluation on the T5 encoder, OpenSora SDDiT, and OpenSora VAE models across diverse datasets reveals that DDiT significantly outperforms state-of-the-art baselines by up to 1.44x in p99 latency and 1.43x in average latency.
title DDiT: Dynamic Resource Allocation for Diffusion Transformer Model Serving
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2506.13497