Utilizing CUDA-Enabled GPUs to Support 5D Scientific Geovisualization: A Case Study of Simulating Dust Storm Events

Li, Jing; Jiang, Yunfeng; Yang, Chaowei; Huang, Qunying

doi:10.1007/978-1-4614-8745-6_6

Cited by 1 publication

(1 citation statement)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…HPC boosts GIS operations and computations in the face of large amounts of geospatial data by utilizing modern hardware architectures such as computer clusters, GPUs, and cloud computing infrastructures [12], [13], [14], [15]. For instance, in [16], GPU accelerators have been used to accelerate the visualization of large-scale geospatial data. In [13], distributed GPU systems through Message Passing Interface (MPI) over Network of Workstations (NoW) and Compute Unified Device Architecture (CUDA) have been used to perform real-time map matching and slope computations of a large global positioning system (GPS) data.…”

Section: Introductionmentioning

confidence: 99%

High Performance Multivariate Geospatial Statistics on Manycore Systems

Salvaña

Abdulah

Huang

et al. 2021

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Modeling and inferring spatial relationships and predicting missing values of environmental data are some of the main tasks of geospatial statisticians. These routine tasks are accomplished using multivariate geospatial models and the cokriging technique. The latter requires the evaluation of the expensive Gaussian log-likelihood function, which has impeded the adoption of multivariate geospatial models for large multivariate spatial datasets. However, this large-scale cokriging challenge provides a fertile ground for supercomputing implementations for the geospatial statistics community as it is paramount to scale computational capability to match the growth in environmental data coming from the widespread use of different data collection technologies. In this paper, we develop and deploy large-scale multivariate spatial modeling and inference on parallel hardware architectures. To tackle the increasing complexity in matrix operations and the massive concurrency in parallel systems, we leverage low-rank matrix approximation techniques with task-based programming models and schedule the asynchronous computational tasks using a dynamic runtime system. The proposed framework provides both the dense and the approximated computations of the Gaussian log-likelihood function. It demonstrates accuracy robustness and performance scalability on a variety of computer systems. Using both synthetic and real datasets, the low-rank matrix approximation shows better performance compared to exact computation, while preserving the application requirements in both parameter estimation and prediction accuracy. We also propose a novel algorithm to assess the prediction accuracy after the online parameter estimation. The algorithm quantifies prediction performance and provides a benchmark for measuring the efficiency and accuracy of several approximation techniques in multivariate spatial modeling.

show abstract