A 28.6 mJ/iter Stable Diffusion Processor for Text-to-Image Generation with Patch Similarity-based Sparsity Augmentation and Text-based Mixed-Precision

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Choi, Jiwon, Jo, Wooyoung, Hong, Seongyon, Kwon, Beomseok, Park, Wonhoon, Yoo, Hoi-Jun
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910615469031424
author Choi, Jiwon
Jo, Wooyoung
Hong, Seongyon
Kwon, Beomseok
Park, Wonhoon
Yoo, Hoi-Jun
author_facet Choi, Jiwon
Jo, Wooyoung
Hong, Seongyon
Kwon, Beomseok
Park, Wonhoon
Yoo, Hoi-Jun
contents This paper presents an energy-efficient stable diffusion processor for text-to-image generation. While stable diffusion attained attention for high-quality image synthesis results, its inherent characteristics hinder its deployment on mobile platforms. The proposed processor achieves high throughput and energy efficiency with three key features as solutions: 1) Patch similarity-based sparsity augmentation (PSSA) to reduce external memory access (EMA) energy of self-attention score by 60.3 %, leading to 37.8 % total EMA energy reduction. 2) Text-based important pixel spotting (TIPS) to allow 44.8 % of the FFN layer workload to be processed with low-precision activation. 3) Dual-mode bit-slice core (DBSC) architecture to enhance energy efficiency in FFN layers by 43.0 %. The proposed processor is implemented in 28 nm CMOS technology and achieves 3.84 TOPS peak throughput with 225.6 mW average power consumption. In sum, 28.6 mJ/iteration highly energy-efficient text-to-image generation processor can be achieved at MS-COCO dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2403_04982
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A 28.6 mJ/iter Stable Diffusion Processor for Text-to-Image Generation with Patch Similarity-based Sparsity Augmentation and Text-based Mixed-Precision
Choi, Jiwon
Jo, Wooyoung
Hong, Seongyon
Kwon, Beomseok
Park, Wonhoon
Yoo, Hoi-Jun
Hardware Architecture
This paper presents an energy-efficient stable diffusion processor for text-to-image generation. While stable diffusion attained attention for high-quality image synthesis results, its inherent characteristics hinder its deployment on mobile platforms. The proposed processor achieves high throughput and energy efficiency with three key features as solutions: 1) Patch similarity-based sparsity augmentation (PSSA) to reduce external memory access (EMA) energy of self-attention score by 60.3 %, leading to 37.8 % total EMA energy reduction. 2) Text-based important pixel spotting (TIPS) to allow 44.8 % of the FFN layer workload to be processed with low-precision activation. 3) Dual-mode bit-slice core (DBSC) architecture to enhance energy efficiency in FFN layers by 43.0 %. The proposed processor is implemented in 28 nm CMOS technology and achieves 3.84 TOPS peak throughput with 225.6 mW average power consumption. In sum, 28.6 mJ/iteration highly energy-efficient text-to-image generation processor can be achieved at MS-COCO dataset.
title A 28.6 mJ/iter Stable Diffusion Processor for Text-to-Image Generation with Patch Similarity-based Sparsity Augmentation and Text-based Mixed-Precision
topic Hardware Architecture
url https://arxiv.org/abs/2403.04982