Exascale Deep Learning for Scientific Inverse Problems

Publication image

We introduce novel communication strategies in synchronous distributed Deep Learning consisting of decentralized gradient reduction orchestration and computational graph-aware grouping of gradient tensors. These new techniques produce an optimal overlap between computation and communication and result in near-linear scaling (0.93) of distributed training up to 27,600 NVIDIA V100 GPUs on the Summit Supercomputer. We demonstrate our gradient reduction techniques in the context of training a Fully Convolutional Neural Network to approximate the solution of a longstanding scientific inverse problem in materials imaging. The efficient distributed training on a dataset size of 0.5 PB, produces a model capable of an atomically-accurate reconstruction of materials, and in the process reaching a peak performance of 2.15(4) EFLOPS (16-bit).


Nouamane Laanait (Oak Ridge National Laboratory)
Joshua Romero (NVIDIA)
Junqi Yin (Oak Ridge National Laboratory)
M. Todd Young (Oak Ridge National Laboratory)
Sean Treichler (NVIDIA)
Vitalii Starchenko (Oak Ridge National Laboratory)
Albina Borisevich (Oak Ridge National Laboratory)
Alex Sergeev (Uber Technologies)
Michael Matheson (Oak Ridge National Laboratory)

Publication Date