arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.09073cs.AR
收藏

DRIFT:利用固有容错性实现高效可靠的扩散模型推理

DRIFT: Harnessing Inherent Fault Tolerance for Efficient and Reliable Diffusion Model Inference

School of Integrated Circuits, Peking University(北京大学集成电路学院);School of Electronics Engineering and Computer Science, Peking University(北京大学电子工程学院);Beijing Advanced Innovation Center for Integrated Circuits(北京集成电路创新中心);Institute of Electronic Design Automation, Peking University(北京大学电子设计自动化研究所)

Jinqi Wen, Tong Xie, Runsheng Wang, Meng Li

AI总结:

本文提出DRIFT框架,通过利用扩散模型的固有容错性,提升推理效率和可靠性,实验显示在电压调低时平均节能36%,超频时提速1.7倍且保持生成质量。

摘要:

尽管在视觉生成任务中表现优异,扩散模型的部署仍面临高能耗和推理延迟问题。动态电压和频率调节(DVFS)提供了一种有前途的解决方案,但现有方法因忽视扩散模型的固有容错性而导致效率提升有限或输出质量下降。因此,本文提出DRIFT,一种新颖的算法架构优化框架,通过利用容错性实现高效可靠的扩散模型推理。我们首先对代表性扩散模型进行了全面的容错性分析。基于这些观察,我们引入了一种细粒度、容错感知的DVFS策略,选择性保护易出错的网络块和时间步,并引入基于回滚算法的容错机制(ABFT),通过回退到先前时间步来适应性纠正关键错误。我们进一步优化卸载间隔并重新组织数据布局以减少内存开销。在多种模型和数据集上的实验表明,DRIFT可通过电压调低实现平均36%的能耗节省,或通过超频实现1.7倍的加速,同时保持生成质量。

英文摘要:

Diffusion model deployment has been suffering from high energy consumption and inference latency despite its superior performance in visual generation tasks. Dynamic voltage and frequency scaling (DVFS) offers a promising solution to exploit the potential of the underlying accelerators. However, existing approaches often lead to either limited efficiency gains or degraded output quality because they overlook the inherent fault tolerance of the diffusion model. Therefore, in this paper, we propose DRIFT, a novel algorithmarchitecture co-optimization framework that harnesses the fault tolerance for efficient and reliable diffusion model inference. We first perform a comprehensive resilience analysis on representative diffusion models. Building on these observations, we introduce a fine-grained, resilience-aware DVFS strategy that selectively protects error-sensitive network blocks and timesteps, and a rollback algorithm-based fault tolerance (ABFT) mechanism that adaptively corrects only critical errors by reverting to previous timesteps. We further optimize offloading intervals and reorganize data layouts to reduce memory overhead. Experiments across diverse models and datasets show that DRIFT can achieve on average 36% energy savings through voltage underscaling or 1.7x speedup via overclocking while maintaining generation 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。

↑