深度学习的兴起促进了各个电子科技行业的迅速发展，军事科学的智能应用也得到了极大的推广。舰载云环境基于单只舰船，为舰载智能应用提供算力、网络资源和存储支持。深度学习任务依赖强大的计算资源，相同的智能模型在不同的计算平台中性能可能出现巨大差异。在自主可控的大背景下，本文基于深度学习中的典型图像识别任务，对比使用国产和商用计算平台中舰载云上不同配置的虚拟机的表现性能。一方面，分析不同的虚拟机配置与表现性能的关系；另一方面，对比商用及国产平台的性能差距，为舰载云智能化的大规模应用提供理论支撑。
The rise of deep learning has rapidly promoted the development of various electronic technology, and the intelligent applications of military science have also been greatly promoted. The shipborne private cloud environment is based on a single ship and provides computing resource, network resources and storage support for shipborne intelligent applications. Deep learning tasks usually rely on powerful computing resources. Even the same model on different platforms may have very different performance. In the context of independence and controllability of the technology, based on the image recognition task in deep learning, this paper compares the performance of virtual machines with different configurations on the shipborne private cloud environment in domestic and commercial platforms. On the one hand, the relationship between performance and different virtual machine configurations is analysed in this paper; on the other hand, the performance of commercial and domestic platforms is compared, which provides theoretical support for the large-scale application of shipborne cloud intelligence.
2021,43(8): 131-134 收稿日期：2020-08-08
DOI：10.3404/j.issn.1672-7649.2021.08.025
分类号：TP332
作者简介：宋达(1993-),男,工程师,主要研究方向:云计算、机器学习
参考文献：
[1] MLPerf[EB/OL]. https: //www.mlperf.org.
[2] 人工智能 深度学习算法评估规范[EB/OL]. http://www.cesi.ac.cn/201807/4058.html.
[3] LECUN Y, BOTTOU L. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278–2324
[4] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]// IEEE Conference on Computer Vision & Pattern Recognition. IEEE Computer Society, 2016.
[5] PASZKE, ADAM et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library[R]. ArXiv abs/1912.01703 (2019): n. pag.
[6] S. SHI, Q. WANG, P. XU, et al. Benchmarking state-of-the-art deep learning software tools[C]// Proceedings of the 7th International Conference on Cloud Computing and Big Data, IEEE, Macau, China, 2016.
