ConvBench:用于二维卷积基元评估的综合基准

Lucas Alvarenga, Victor Ferrari, Rafael Souza, Marcio Pereira, Guido Araujo
{"title":"ConvBench:用于二维卷积基元评估的综合基准","authors":"Lucas Alvarenga, Victor Ferrari, Rafael Souza, Marcio Pereira, Guido Araujo","doi":"arxiv-2407.10730","DOIUrl":null,"url":null,"abstract":"Convolution is a compute-intensive operation placed at the heart of\nConvolution Neural Networks (CNNs). It has led to the development of many\nhigh-performance algorithms, such as Im2col-GEMM, Winograd, and\nDirect-Convolution. However, the comparison of different convolution algorithms\nis an error-prone task as it requires specific data layouts and system\nresources. Failure to address these requirements might lead to unwanted time\npenalties. Thus, considering all processing steps within convolution algorithms\nis essential to comprehensively evaluate and fairly compare their performance.\nFurthermore, most known convolution benchmarking adopts ad-hoc testing suites\nwith limited coverage and handmade operations. This paper proposes ConvBench, a\nprimitive-level benchmark for the evaluation and comparison of convolution\nalgorithms. It assesses 9243 convolution operations derived from 1097\nreal-world deep learning models, resulting in performance and execution\nbreakdown graphs for a detailed evaluation. ConvBench capability is evaluated\nacross the Sliced Convolution (SConv) algorithm. The experiments showed results\nfaster than Im2col-GEMM in 93.6% of the convolutions. However, the use of\nConvBench allowed the delving into the remaining 6.4% underperforming\nconvolutions, uncovering a critical slowdown of 79.5% on average of SConv's\npacking step. This analysis underscores a potential source of optimization for\nSConv, opening up new paths for convolution designers to improve their\nalgorithms.","PeriodicalId":501291,"journal":{"name":"arXiv - CS - Performance","volume":"36 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"ConvBench: A Comprehensive Benchmark for 2D Convolution Primitive Evaluation\",\"authors\":\"Lucas Alvarenga, Victor Ferrari, Rafael Souza, Marcio Pereira, Guido Araujo\",\"doi\":\"arxiv-2407.10730\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Convolution is a compute-intensive operation placed at the heart of\\nConvolution Neural Networks (CNNs). It has led to the development of many\\nhigh-performance algorithms, such as Im2col-GEMM, Winograd, and\\nDirect-Convolution. However, the comparison of different convolution algorithms\\nis an error-prone task as it requires specific data layouts and system\\nresources. Failure to address these requirements might lead to unwanted time\\npenalties. Thus, considering all processing steps within convolution algorithms\\nis essential to comprehensively evaluate and fairly compare their performance.\\nFurthermore, most known convolution benchmarking adopts ad-hoc testing suites\\nwith limited coverage and handmade operations. This paper proposes ConvBench, a\\nprimitive-level benchmark for the evaluation and comparison of convolution\\nalgorithms. It assesses 9243 convolution operations derived from 1097\\nreal-world deep learning models, resulting in performance and execution\\nbreakdown graphs for a detailed evaluation. ConvBench capability is evaluated\\nacross the Sliced Convolution (SConv) algorithm. The experiments showed results\\nfaster than Im2col-GEMM in 93.6% of the convolutions. However, the use of\\nConvBench allowed the delving into the remaining 6.4% underperforming\\nconvolutions, uncovering a critical slowdown of 79.5% on average of SConv's\\npacking step. This analysis underscores a potential source of optimization for\\nSConv, opening up new paths for convolution designers to improve their\\nalgorithms.\",\"PeriodicalId\":501291,\"journal\":{\"name\":\"arXiv - CS - Performance\",\"volume\":\"36 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-07-15\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Performance\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2407.10730\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Performance","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2407.10730","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

卷积是一种计算密集型操作,是卷积神经网络(CNN)的核心。它催生了许多高性能算法,如 Im2col-GEMM、Winograd 和直接卷积。然而,比较不同的卷积算法是一项容易出错的任务,因为它需要特定的数据布局和系统资源。如果不能满足这些要求,可能会导致不必要的时间损失。因此,考虑卷积算法中的所有处理步骤对于全面评估和公平比较它们的性能至关重要。此外,大多数已知的卷积基准测试都采用了覆盖范围有限的临时测试套件和手工操作。本文提出了用于评估和比较卷积算法的原始级基准 ConvBench。它评估了来自 1097 个真实世界深度学习模型的 9243 个卷积操作,并绘制了性能和执行分解图以进行详细评估。ConvBench 评估了切片卷积(SConv)算法的能力。实验显示,在 93.6% 的卷积中,ConvBench 的结果比 Im2col-GEMM 更快。不过,使用 ConvBench 可以深入研究其余 6.4% 表现不佳的卷积,发现 SConv 的堆积步骤平均减慢了 79.5% 的关键速度。这项分析强调了 SConv 的潜在优化来源,为卷积设计人员改进算法开辟了新的途径。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
ConvBench: A Comprehensive Benchmark for 2D Convolution Primitive Evaluation
Convolution is a compute-intensive operation placed at the heart of Convolution Neural Networks (CNNs). It has led to the development of many high-performance algorithms, such as Im2col-GEMM, Winograd, and Direct-Convolution. However, the comparison of different convolution algorithms is an error-prone task as it requires specific data layouts and system resources. Failure to address these requirements might lead to unwanted time penalties. Thus, considering all processing steps within convolution algorithms is essential to comprehensively evaluate and fairly compare their performance. Furthermore, most known convolution benchmarking adopts ad-hoc testing suites with limited coverage and handmade operations. This paper proposes ConvBench, a primitive-level benchmark for the evaluation and comparison of convolution algorithms. It assesses 9243 convolution operations derived from 1097 real-world deep learning models, resulting in performance and execution breakdown graphs for a detailed evaluation. ConvBench capability is evaluated across the Sliced Convolution (SConv) algorithm. The experiments showed results faster than Im2col-GEMM in 93.6% of the convolutions. However, the use of ConvBench allowed the delving into the remaining 6.4% underperforming convolutions, uncovering a critical slowdown of 79.5% on average of SConv's packing step. This analysis underscores a potential source of optimization for SConv, opening up new paths for convolution designers to improve their algorithms.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
HRA: A Multi-Criteria Framework for Ranking Metaheuristic Optimization Algorithms Temporal Load Imbalance on Ondes3D Seismic Simulator for Different Multicore Architectures Can Graph Reordering Speed Up Graph Neural Network Training? An Experimental Study The Landscape of GPU-Centric Communication A Global Perspective on the Past, Present, and Future of Video Streaming over Starlink
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1