MLPerf Training Benchmark
Abstract
Machine learning (ML) needs industry-standard performance benchmarks to support design and competitive evaluation of the many emerging software and hardware solutions for ML. But ML training presents three unique benchmarking challenges absent from other domains: optimizations that improve training throughput can increase the time to solution, training is stochastic and time to solution exhibits high variance, and software and hardware systems are so diverse that fair benchmarking with the same binary, code, and even hyperparameters is difficult. We therefore present MLPerf, an ML benchmark that overcomes these challenges. Our analysis quantitatively evaluates MLPerf's efficacy at driving performance and scalability improvements across two rounds of results from multiple vendors.
Cite
@article{arxiv.1910.01500,
title = {MLPerf Training Benchmark},
author = {Peter Mattson and Christine Cheng and Cody Coleman and Greg Diamos and Paulius Micikevicius and David Patterson and Hanlin Tang and Gu-Yeon Wei and Peter Bailis and Victor Bittorf and David Brooks and Dehao Chen and Debojyoti Dutta and Udit Gupta and Kim Hazelwood and Andrew Hock and Xinyuan Huang and Atsushi Ike and Bill Jia and Daniel Kang and David Kanter and Naveen Kumar and Jeffery Liao and Guokai Ma and Deepak Narayanan and Tayo Oguntebi and Gennady Pekhimenko and Lillian Pentecost and Vijay Janapa Reddi and Taylor Robie and Tom St. John and Tsuguchika Tabaru and Carole-Jean Wu and Lingjie Xu and Masafumi Yamazaki and Cliff Young and Matei Zaharia},
journal= {arXiv preprint arXiv:1910.01500},
year = {2020}
}
Comments
MLSys 2020