This work is a reproduction of ParaDnn (originally implemented with TensorFlow), a benchmark set of hyper-parameterized DNNs. We rewrite the benchmarks with Pytorch. The bibtex information of the original paper is as below.
@inproceedings{wang2020systematic,
title={A Systematic Methodology for Analysis of Deep Learning Hardware and Software Platforms},
author={Wang, Yu Emma and Wei, Gu-Yeon and Brooks, David},
booktitle={The 3rd Conference on Machine Learning and Systems (MLSys)},
year={2020}
}
We test our benchmark set on two JLab ifarm GPUs: NVIDIA T4 and A100 80GB PCIe of compute capacity 7.5 and 8.0, respectively. Prelimary results are presented in the CHEP 2023 poster.