Acceleration of Graph Neural Network-Based Prediction Models in Chemistry via Co-Design Optimization on Intelligence Processing Units

Atomic structure prediction and associated property calculations are the bedrock of chemical physics. Since high-fidelity ab initio modeling techniques for computing the structure and properties can be prohibitively expensive, this motivates the development of machine-learning (ML) models that make these predictions more efficiently. Training graph neural networks over large atomistic databases introduces unique computational challenges, such as the need to process millions of small graphs with variable size and support communication patterns that are distinct from learning over large graphs, such as social networks. We demonstrate a novel hardware-software codesign approach to scale up the training of atomistic graph neural networks (GNN) for structure and property prediction. First, to eliminate redundant computation and memory associated with alternative padding techniques and to improve throughput via minimizing communication, we formulate the effective coalescing of the batches of variable-size atomistic graphs as the bin packing problem and introduce a hardware-agnostic algorithm to pack these batches. In addition, we propose hardware-specific optimizations, including a planner and vectorization for the gather-scatter operations targeted for Graphcore's Intelligence Processing Unit (IPU), as well as model-specific optimizations such as merged communication collectives and optimized softplus. Putting these all together, we demonstrate the effectiveness of the proposed codesign approach by providing an implementation of a well-established atomistic GNN on the Graphcore IPUs. We evaluate the training performance on multiple atomistic graph databases with varying degrees of graph counts, sizes, and sparsity. We demonstrate that such a codesign approach can reduce the training time of atomistic GNNs and can improve their performance by up to 1.5× compared to the baseline implementation of the model on the IPUs. Additionally, we compare our IPU implementation with a Nvidia GPU-based implementation and show that our atomistic GNN implementation on the IPUs can run 1.8× faster on average compared to the execution time on the GPUs.

Electronic structure simulations in the cloud computing environment

The transformative impact of modern computational paradigms and technologies, such as high-performance computing (HPC), quantum computing, and cloud computing, has opened up profound new opportunities for scientific simulations. Scalable computational chemistry is one beneficiary of this technological progress. The main focus of this paper is on the performance of various quantum chemical formulations, ranging from low-order methods to high-accuracy approaches, implemented in different computational chemistry packages and libraries, such as NWChem, NWChemEx, Scalable Predictive Methods for Excitations and Correlated Phenomena, ExaChem, and Fermi-Löwdin orbital self-interaction correction on Azure Quantum Elements, Microsoft's cloud services platform for scientific discovery. We pay particular attention to the intricate workflows for performing complex chemistry simulations, associated data curation, and mechanisms for accuracy assessment, which is demonstrated with the Arrows automated workflow for high throughput simulations. Finally, we provide a perspective on the role of cloud computing in supporting the mission of leadership computational facilities.