arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2501.09253cs.DC
收藏

PATCHEDSERVE:面向 SLO 优化的混合分辨率扩散服务的补丁管理框架

PATCHEDSERVE: A Patch Management Framework for SLO-Optimized Hybrid Resolution Diffusion Serving

Desen Sun, Zepeng Zhao, Yuke Wang

AI总结:

针对多分辨率 T2I 扩散服务批处理与缓存难题,提出 PatchedServe,通过补丁化工作流、补丁级缓存复用和 SLO 感知调度提升服务质量。

摘要:

文本到图像(Text-to-Image,T2I)扩散模型已成为应用最广泛的生成模型之一。然而,以整幅图像为粒度提供扩散模型服务会带来显著挑战,尤其是在多分辨率工作负载下。首先,图像级服务阻碍了跨请求的批处理。其次,异构分辨率表现出不同的局部性特征,使得难以有效地应用统一的缓存策略。为应对这些挑战,我们提出了 PatchedServe,这是一个用于 SLO 优化的混合分辨率扩散服务的补丁管理框架。PatchedServe 是首个旨在处理异构分辨率、经过 SLO 优化的 T2I 扩散服务框架。具体而言,它引入了一种新颖的基于补丁的处理工作流,可显著提升混合分辨率输入的吞吐量。此外,PatchedServe 设计了补丁级缓存复用策略,以充分利用扩散计算中的冗余,并集成了具备轻量级在线延迟预测、感知 SLO 的调度算法来提升响应能力。我们的评估表明,在保持图像质量的同时,PatchedServe 的 SLO 满足率比最先进的扩散服务系统高出 30.1%。

英文摘要:

The Text-to-Image (T2I) diffusion model has emerged as one of the most widely adopted generative models. However, serving diffusion models at the granularity of entire images introduces significant challenges, particularly under multi-resolution workloads. First, image-level serving obstructs batching across requests. Second, heterogeneous resolutions exhibit distinct locality characteristics, making it difficult to apply a uniform cache policy effectively. To address these challenges, we present PatchedServe, a Patch Management Framework for SLO-Optimized Hybrid-Resolution Diffusion Serving. PatchedServe is the first SLO-optimized T2I diffusion serving framework designed to handle heterogeneous resolutions. Specifically, it incorporates a novel patch-based processing workflow that substantially improves throughput for hybrid-resolution inputs. Moreover, PatchedServe devises a patch-level cache reuse policy to fully exploit diffusion redundancies and integrates an SLO-aware scheduling algorithm with lightweight online latency prediction to improve responsiveness. Our evaluation demonstrates that PatchedServe achieves 30.1 % higher SLO satisfaction than the state-of-the-art diffusion serving system, while preserving image quality.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑