<span class="var-sub_title">CosmoFlow: Using Deep Learning to Learn the Universe at Scale</span> SC18 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

CosmoFlow: Using Deep Learning to Learn the Universe at Scale

Authors: Amrita Mathuriya (Intel Corporation), Deborah Bard (National Energy Research Scientific Computing Center (NERSC), Lawrence Berkeley National Laboratory), Pete Mendygral (Cray Inc), Lawrence Meadows (Intel Corporation), James Arnemann (University of California, Berkeley), Lei Shao (Intel Corporation), Siyu He (Carnegie Mellon University), Tuomas Karna (Intel Corporation), Diana Moise (Cray Inc), Simon J. Pennycook (Intel Corporation), Kristyn Maschhoff (Cray Inc), Jason Sewall (Intel Corporation), Nalini Kumar (Intel Corporation), Shirley Ho (Lawrence Berkeley National Laboratory, Carnegie Mellon University), Michael F. Ringenburg (Cray Inc), Mr Prabhat (Lawrence Berkeley National Laboratory, National Energy Research Scientific Computing Center (NERSC)), Victor Lee (Intel Corporation)

Abstract: Deep learning is a promising tool to determine the physical model that describes our universe. To handle the considerable computational cost of this problem, we present CosmoFlow: a highly scalable deep learning application built on top of the TensorFlow framework.

CosmoFlow uses efficient implementations of 3D convolution and pooling primitives, together with improvements in threading for many element-wise operations, to improve training performance on Intel Xeon Phi processors. We also utilize the Cray PE Machine Learning Plugin for efficient scaling to multiple nodes. We demonstrate fully synchronous data-parallel training on 8192 nodes of Cori with 77% parallel efficiency, achieving 3.5 Pflop/s sustained performance.

To our knowledge, this is the first large-scale science application of the TensorFlow framework at supercomputer scale with fully-synchronous training. These enhancements enable us to process large 3D dark matter distribution and predict the cosmological parameters Omega_M, sigma_8 and N_s with unprecedented accuracy.

Back to Technical Papers Archive Listing