一句话：提出步级序列并行与轮次调度，提升混合 DiT 负载的 SLO 达成率与 GPU 利用率。
扩散 Transformer（DiT）通过迭代去噪生成高质量图像，但在严格 SLO 下服务计算开销大，尤其高分辨率场景。现有系统采用固定度序列并行，难以应对混合分辨率与截止时间的异构负载，导致 GPU 利用率低、SLO 达成率差。TetriServe 提出步级序列并行，按请求截止时间动态调整并行度；并设计轮次调度机制，将时间离散为固定轮次，使截止时间感知调度可行，同时在步级别适配并行度、最小化 GPU 小时消耗，并联合打包请求以减少延迟完成。在先进 DiT 模型上评估显示，SLO 达成率最高提升 32%，且图像质量不下降。
Diffusion Transformer (DiT) models excel at generating high-quality images through iterative denoising steps, but serving them under strict Service Level Objectives (SLOs) is challenging due to their high computational cost, particularly at larger resolutions. Existing serving systems use fixed-degree sequence parallelism, which is inefficient for heterogeneous workloads with mixed resolutions and deadlines, leading to poor GPU utilization and low SLO attainment. In this paper, we propose step-level sequence parallelism to dynamically adjust the degree of parallelism of individual requests according to their deadlines. We present TetriServe1, a DiT serving system that implements this strategy for highly efficient image generation. Specifically, TetriServe introduces a novel round-based scheduling mechanism that improves SLO attainment by (1) discretizing time into fixed rounds to make deadline-aware scheduling tractable, (2) adapting parallelism at the step level and minimizing GPU hour consumption, and (3) jointly packing requests to minimize late completions. Extensive evaluation on state-of-the-art DiT models shows that TetriServe achieves up to 32% higher SLO attainment compared to existing solutions without degrading image quality. TetriServe is available at https://github.com/DiT-Serving/TetriServe.
DOI 原文 · arXiv · PDF（开放获取） · DBLP
@inproceedings{LuHTLWM0C26,
title = {TetriServe: Efficiently Serving Mixed DiT Workloads},
author = {Runyu Lu and Shiqi He and Wenxuan Tan and Shenggui Li and Ruofan Wu and Jeff J. Ma and Ang Chen and Mosharaf Chowdhury},
booktitle = {ASPLOS},
year = {2026},
doi = {10.1145/3779212.3790233}
}