GenCNN: A Partition-Aware Multi-Objective Mapping Framework for CNN Accelerators Based on Genetic Algorithm

doi:10.1145/3747844

	GenCNN: A Partition-Aware Multi-Objective Mapping Framework for CNN Accelerators Based on Genetic Algorithm
	Mu, Yudong 1,2; Fan, Zhihua 1,2; Li, Wenming 1,2; Zhang, Zhiyuan 1,2; An, Xuejun 1; Fan, Dongrui 1,2; Ye, Xiaochun 1,2
	2025-09-01
发表期刊	ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION
ISSN	1544-3566
卷号	22 期号:3 页码:26
摘要	Convolutional Neural Networks (CNNs) require partitioning to efficiently run on CNN accelerators, which offer multiple parallel processing dimensions, such as Processing Element (PE) array topologies and Single Instruction Multiple Data (SIMD) execution. The choice of parallelization strategy directly impacts accelerator performance. However, the vast search space for CNN partitioning and parallelization makes manual optimization costly and complex, especially when addressing both aspects simultaneously. This highlights the need for an automated framework to efficiently map CNNs onto accelerators. Our key insight is that existing approaches suffer from inadequate accelerator performance modeling and a lack of multi-objective optimization strategies that jointly consider task partitioning and convolution parallelization. To address this, we propose GenCNN, a multi-objective genetic algorithm-based mapping framework for CNN accelerators. GenCNN first constructs a fine-grained performance model that captures both off-chip data access and on-chip data processing. It then applies the Non-dominated Sorting Genetic Algorithm II improved by Multi-Objective Bayesian Optimization to derive a Pareto-optimal partitioning and parallelization strategy that balances off-chip latency and PE utilization. Finally, GenCNN optimizes scheduling and routing to minimize data transfers. Experimental results show that GenCNN achieves up to 17.66x speedup in compilation and 6.47x in execution compared with state-of-the-art mapping frameworks.
关键词	CNN Accelerator Dataflow Graph Mapping Genetic Algorithm Multi-objective Optimization
DOI	10.1145/3747844
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2023YFB4503500] ; Beijing Nova Program[20220484054] ; Beijing Nova Program[20230484420] ; Beijing Natural Science Foundation[L234078] ; SKLP Foundation[CLQD202502]
WOS研究方向	Computer Science
WOS类目	Computer Science, Hardware & Architecture ; Computer Science, Theory & Methods
WOS记录号	WOS:001606458500004
出版者	ASSOC COMPUTING MACHINERY
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/41577
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Fan, Zhihua; Li, Wenming
作者单位	1.Chinese Acad Sci, Inst Comp Technol, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp Sci & Technol, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Mu, Yudong,Fan, Zhihua,Li, Wenming,et al. GenCNN: A Partition-Aware Multi-Objective Mapping Framework for CNN Accelerators Based on Genetic Algorithm[J]. ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,2025,22(3):26.
APA	Mu, Yudong.,Fan, Zhihua.,Li, Wenming.,Zhang, Zhiyuan.,An, Xuejun.,...&Ye, Xiaochun.(2025).GenCNN: A Partition-Aware Multi-Objective Mapping Framework for CNN Accelerators Based on Genetic Algorithm.ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,22(3),26.
MLA	Mu, Yudong,et al."GenCNN: A Partition-Aware Multi-Objective Mapping Framework for CNN Accelerators Based on Genetic Algorithm".ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION 22.3(2025):26.