Speed Is All You Need: On-Device Acceleration of Large Diffusion Models via GPU-Aware Optimizations

Author:

Chen Yu-Hui1,Sarokin Raman1,Lee Juhyun1,Tang Jiuqiang1,Chang Chuo-Ling1,Kulik Andrei1,Grundmann Matthias1

Affiliation:

1. Google LLC,Mountain View,CA,94043

Publisher

IEEE

Reference26 articles.

1. A style-based generator architecture for generative adversarial networks;karras;CoRR,2018

2. World’s first on-device demonstration of stable diffusion on an android phone;hou,2023

3. Reformer: The efficient transformer;kitaev;CoRR,2020

4. Denoising diffusion probabilistic models;ho;CoRR,2020

Cited by 6 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. SDAcc: A Stable Diffusion Accelerator on FPGA via Software-Hardware Co-Design;2024 IEEE 32nd Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM);2024-05-05

2. SmartMem: Layout Transformation Elimination and Adaptation for Efficient DNN Execution on Mobile;Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3;2024-04-27

3. Accelerator-Aware Computation Offloading Under Timing Constraints;2024 International Conference on Computing, Networking and Communications (ICNC);2024-02-19

4. Towards Using Generative AI for Facilitating Image Creation in Spatial Augmented Reality;2023 IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct);2023-10-16

5. Softmax Bias Correction for Quantized Generative Models;2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW);2023-10-02

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3