arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2310.09554cs.LG
收藏

神经网络评分以实现高效计算

Neural network scoring for efficient computing

SONDRA;CentraleSupélec(巴黎中央理工-高等电力学院);Université Paris Saclay(巴黎萨克雷大学);CNRS(法国国家科学研究中心);Sorbonne Université(索邦大学)

Hugo Waltsburger, Erwan Libessart, Chengfang Ren, Anthony Kolar, Regis Guinvarc'h

AI总结:

本文提出一种综合评分工具,用于测量神经网络推理中的功耗与准确率权衡,支持多指标评估,首次实现硬件架构适配性测试,以优化算法效率。

摘要:

大量工作致力于估计和优化高性能计算(HPC)与深度学习中的工作负载。然而,研究人员通常依赖少数几个指标来评估这些技术的效率,最显著的是预测的准确率、损失以及相对于GPU和/或CPU特性的计算时间。功耗数据很少被展示,部分原因在于难以获得准确的功耗读数。在本文中,我们引入了一个综合评分,旨在刻画神经网络推理过程中测量到的准确率与功耗之间的权衡。为此,我们推出了一款新的开源工具,使研究人员能够考虑更多指标:细粒度的功耗,以及RAM/CPU/GPU利用率、存储和网络输入/输出(I/O)。据我们所知,这是首个针对硬件架构的神经网络适配性测试。这得益于可复现的功耗效率测量。我们将此流程应用于多种硬件上的最先进神经网络架构。主要应用和新颖之处之一在于算法功耗效率的测量,目标是让研究人员更好地掌握其算法的效率。该方法的开发旨在探索神经网络中能耗与准确率之间的权衡,同时在进行架构探索时,它也适用于为特定任务适配硬件或更准确地比较两种架构。

英文摘要:

Much work has been dedicated to estimating and optimizing workloads in high-performance computing (HPC) and deep learning. However, researchers have typically relied on few metrics to assess the efficiency of those techniques. Most notably, the accuracy, the loss of the prediction, and the computational time with regard to GPUs or/and CPUs characteristics. It is rare to see figures for power consumption, partly due to the difficulty of obtaining accurate power readings. In this paper, we introduce a composite score that aims to characterize the trade-off between accuracy and power consumption measured during the inference of neural networks. For this purpose, we present a new open-source tool allowing researchers to consider more metrics: granular power consumption, but also RAM/CPU/GPU utilization, as well as storage, and network input/output (I/O). To our best knowledge, it is the first fit test for neural architectures on hardware architectures. This is made possible thanks to reproducible power efficiency measurements. We applied this procedure to state-of-the-art neural network architectures on miscellaneous hardware. One of the main applications and novelties is the measurement of algorithmic power efficiency. The objective is to allow researchers to grasp their algorithms' efficiencies better. This methodology was developed to explore trade-offs between energy usage and accuracy in neural networks. It is also useful when fitting hardware for a specific task or to compare two architectures more accurately, with architecture exploration in mind.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑