CosmoFlow: Using Deep Learning to Learn the Universe at Scale

TL;DR

CosmoFlow leverages scalable deep learning on supercomputers to analyze large-scale cosmological data, achieving high efficiency and accuracy in predicting universe parameters.

Contribution

This work introduces a highly scalable TensorFlow-based deep learning application for cosmology, optimized for supercomputing environments and demonstrating unprecedented performance.

Findings

Achieved 3.5 Pflop/s performance on 8192 nodes

Demonstrated fully synchronous training with 77% efficiency

Predicted cosmological parameters with unprecedented accuracy

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(C) Xeon Phi(TM) 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.