A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning
Authors:
- 1 May 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 66-77
Abstract
We introduce Deep500: the first customizable benchmarking infrastructure that enables fair comparison of the plethora of deep learning frameworks, algorithms, libraries, and techniques. The key idea behind Deep500 is its modular design, where deep learning is factorized into four distinct levels: operators, network processing, training, and distributed training. Our evaluation illustrates that Deep500 is customizable (enables combining and benchmarking different deep learning codes) and fair (uses carefully selected metrics). Moreover, Deep500 is fast (incurs negligible overheads), verifiable (offers infrastructure to analyze correctness), and reproducible. Finally, as the first distributed and reproducible benchmarking system for deep learning, Deep500 provides software infrastructure to utilize the most powerful supercomputers for extreme-scale workloads.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- Evolving Deep Networks Using HPC2017
- TensorLayer2017
- Fathom: reference workloads for modern deep learning methods2016
- Deep Residual Learning for Image Recognition2016
- Fast Algorithms for Convolutional Neural Networks2016
- Wide Residual Networks2016
- Scientific benchmarking of parallel computing systems2015
- Caffe2014
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998
- A Stochastic Approximation MethodThe Annals of Mathematical Statistics, 1951