Abstract
Graph neural networks (GNNs) are a class of Deep Learning models used in designing atomistic materials for effective screening of large chemical spaces. To ensure robust prediction, GNN models must be trained on large volumes of atomistic data on leadership class supercomputers. Even with the advent of modern architectures that consist of multiple storage layers that include node-local NVMe devices in addition to device memory for caching large datasets, extreme-scale model training faces I/O challenges at scale. We present DDStore, an in-memory distributed data store designed for GNN training on large-scale graph data. DDStore provides a hierarchical, distributed, data caching technique that combines data chunking, replication, low-latency random access, and high throughput communication. DDStore achieves near-linear scaling for training a GNN model using up to 1000 GPUs on the Summit and Perlmutter supercomputers, and reaches up to a 6.15x reduction in GNN training time compared to state-of-the-art methodologies.
Original language | English |
---|---|
Title of host publication | Proceedings of 2023 SC Workshops of the International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 |
Publisher | Association for Computing Machinery |
Pages | 941-950 |
Number of pages | 10 |
ISBN (Electronic) | 9798400707858 |
DOIs | |
State | Published - Nov 12 2023 |
Event | 2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 - Denver, United States Duration: Nov 12 2023 → Nov 17 2023 |
Publication series
Name | ACM International Conference Proceeding Series |
---|
Conference
Conference | 2023 International Conference on High Performance Computing, Network, Storage, and Analysis, SC Workshops 2023 |
---|---|
Country/Territory | United States |
City | Denver |
Period | 11/12/23 → 11/17/23 |
Funding
*This manuscript has been authored in part by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a non-exclusive, paid up, irrevocable, world-wide license to publish or reproduce the published form of the manuscript, or allow others to do so, for U.S. Government purposes. The DOE will provide public access to these results in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). This research is sponsored by the Artificial Intelligence Initiative as part of the Laboratory Directed Research and Development (LDRD) Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US Department of Energy under contract DE-AC05-00OR22725. This work has been supported by the SciDAC Institute for Computer Science, Data, and Artificial Intelligence (RAPIDS), Lawrence Berkeley National Laboratory, which is operated by the University of California for the U.S. Department of Energy under contract DE-AC02-05CH11231. This research used resources of the Oak Ridge Leadership Computing Facility at the Oak Ridge National Laboratory and the National Energy Research Scientific Computing Center (NERSC), which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC05-00OR22725 and No. DE-AC02-05CH11231 using NERSC award ASCR-ERCAP0025216, respectively.
Keywords
- Atomistic Modeling
- Deep Learning
- Distributed Data Parallelism
- Graph Neural Networks
- Inorganic Chemistry
- Organic Chemistry
- Quantum Chemistry