Applying the Roofline model for Deep Learning performance optimizations

Autor:	Czaja, Jacek, Gallus, Michal, Wozna, Joanna, Grygielski, Adam, Tao, Luo
Rok vydání:	2020
Předmět:	Computer Science - Distributed Parallel and Cluster Computing Computer Science - Artificial Intelligence Computer Science - Performance
Druh dokumentu:	Working Paper
Popis:	In this paper We present a methodology for creating Roofline models automatically for Non-Unified Memory Access (NUMA) using Intel Xeon as an example. Finally, we present an evaluation of highly efficient deep learning primitives as implemented in the Intel oneDNN Library. Comment: oneDNN library analysis with roofline model
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2009.11224 Zobrazit plný text záznamu View this record from Arxiv