Fast Stencil-Code Computation on a Wafer-Scale Processor

Rocki, Kamil; Essendelft, Dirk Van; Sharapov, Ilya; Schreiber, Robert D.; Morrison, Michael L.; Kibardin, Vladimir; Portnoy, Andrey; Dietiker, Jean François; Syamlal, Madhava; James, Michael

doi:10.48550/arxiv.2010.03660

Cited by 2 publications

(2 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the cyber-infrastructure side, next-generation AI/ML acceleration hardware continuously evolve to tackle the scalability issue. For example, a recent pilot study in computational fluid dynamics showed that it could be more than 200 times faster than the same workload on an optimized number of cores on the NETL's supercomputer JOULE 2.0 [Rocki et al, 2020]. Similar scaling performance has been reported on other exascale computing clusters involving hundreds of GPU's [Byna et al, 2020].…”

Section: Model Scalibilitymentioning

confidence: 70%

Applications of physics-informed scientific machine learning in subsurface science: A survey

Sun¹,

Yoon²,

Shih³

et al. 2021

Preprint

View full text Add to dashboard Cite

Geosystems are geological formations altered by humans activities such as fossil energy exploration, waste disposal, geologic carbon sequestration, and renewable energy generation. Geosystems also represent a critical link in the global water-energy nexus, providing both the source and buffering mechanisms for enabling societal adaptation to climate variability and change. The responsible use and exploration of geosystems are thus critical to the geosystem governance, which in turn depends on the efficient monitoring, risk assessment, and decision support tools for practical implementation. Large-scale, physics-based models have long been developed and used for geosystem management by incorporating geological domain knowledge such as stratigraphy, governing equations of flow and mass transport in porous media, geological and initial/boundary constraints, and field observations. Spatial heterogeneities and the multiscale nature of geological formations, however, pose significant challenges to the conventional numerical models, especially when used in a simulation-based optimization framework for decision support. Fast advances in machine learning (ML) algorithms and novel sensing technologies in recent years have presented new opportunities for the subsurface research community to improve the efficacy and transparency of geosystem governance. Although recent studies have shown the great promise of scientific ML (SciML) models, questions remain on how to best leverage ML in the management of geosystems, which are typified by multiscality, high-dimensionality, and data resolution inhomogeneity. This survey will provide a systematic review of the recent development and applications of domain-aware SciML in geosystem researches, with an emphasis on how the accuracy, interpretability, scalability, defensibility, and generalization skill of ML approaches can be improved to better serve the geoscientific community.

show abstract

Section: Model Scalibilitymentioning

confidence: 70%

Applications of physics-informed scientific machine learning in subsurface science: A survey

Sun¹,

Yoon²,

Shih³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…For example, the Lassen system at Lawrence Livermore National Labs has a Cerebras accelerator integrated with it. Recent results show a 0.86 PFLOPS on a single wafer scale chip [8] on stencil problems. Graphcore IPUs and SambaNova are starting to be use in traditional HPC applications [9], [10].…”

Section: Introductionmentioning

confidence: 99%

Evaluating Spatial Accelerator Architectures with Tiled Matrix-Matrix Multiplication

Moon¹,

Kwon²,

Jeong³

et al. 2021

Preprint

View full text Add to dashboard Cite

There is a growing interest in custom spatial accelerators for machine learning applications. These accelerators employ a spatial array of processing elements (PEs) interacting via custom buffer hierarchies and networks-on-chip. The efficiency of these accelerators comes from employing optimized dataflow (i.e., spatial/temporal partitioning of data across the PEs and fine-grained scheduling) strategies to optimize data reuse. The focus of this work is to evaluate these accelerator architectures using a tiled general matrix-matrix multiplication (GEMM) kernel. To do so, we develop a framework that finds optimized mappings (dataflow and tile sizes) for a tiled GEMM for a given spatial accelerator and workload combination, leveraging an analytical cost model for runtime and energy. Our evaluations over five spatial accelerators demonstrate that the tiled GEMM mappings systematically generated by our framework achieve high performance on various GEMM workloads and accelerators.

show abstract

Fast Stencil-Code Computation on a Wafer-Scale Processor

Cited by 2 publications

References 13 publications

Applications of physics-informed scientific machine learning in subsurface science: A survey

Applications of physics-informed scientific machine learning in subsurface science: A survey

Evaluating Spatial Accelerator Architectures with Tiled Matrix-Matrix Multiplication

Contact Info

Product

Resources

About