Analytical characterization and design space exploration for optimization of CNNs
Analytical characterization and design space exploration for optimization of CNNs
复制标题
用于优化 CNN 的分析表征和设计空间探索
DOI：
10.1145/3445814.3446759
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Sadayappan, P.
中科院分区：
文献类型：
--
作者：
Li, Rui;Xu, Yufan;Sukumaran-Rajam, Aravind;Rountev, Atanas;Sadayappan, P.
通过内存层次结构移动数据是一个基本的瓶颈，可能会限制机器学习核心算法的性能，例如卷积神经网络（CNN）。循环级优化，包括循环平铺和循环置换，是减少数据移动的基本转换。然而，寻找最佳循环级优化配置的搜索空间是爆炸性的大。本文开发了一种分析建模方法，用于在多核CPU上找到CNN的最佳循环级优化配置。实验评估表明，这种方法实现了与最先进的库和基于自动调整的CNN优化器相当或更好的性能。
Moving data through the memory hierarchy is a fundamental bottleneck that can limit the performance of core algorithms of machine learning, such as convolutional neural networks (CNNs). Loop-level optimization, including loop tiling and loop permutation, are fundamental transformations to reduce data movement. However, the search space for finding the best loop-level optimization configuration is explosively large. This paper develops an analytical modeling approach for finding the best loop-level optimization configuration for CNNs on multi-core CPUs. Experimental evaluation shows that this approach achieves comparable or better performance than state-of-the-art libraries and auto-tuning based optimizers for CNNs.