Distributed Sparse Block Grids on GPUs.

First Authors Pietro Incardona
Authors Pietro Incardona, Tommaso Bianucci, Ivo F. Sbalzarini
Corresponding Authors Ivo F. Sbalzarini
Last Authors Ivo F. Sbalzarini
Conference Proceedings Volume Title High Performance Computing : 36th International Conference, ISC High Performance 2021, Virtual Event, June 24 – July 2, 2021, Proceedings
Series Title (Lecture Notes in Computer Science ; 12728)
Conference Name International Conference on High Performance Computing (ISC)
Conference Location online
Conference Start Date 2021-06-24
Conference End Date 2021-07-02
Chapter Number
Publisher Springer International Publishing
Conference Proceedings Editors
ISBN 978-3-030-78713-4
First Page 272
Last Page 290
Open Access false
Print Publication Date 2021-07-02
Online Publication Date 2021-07-02
Abstract We present a design and implementation of distributed sparse block grids that transparently scale from a single CPU to multi-GPU clusters. We support dynamic sparse grids as, e.g., occur in computer graphics with complex deforming geometries and in multi-resolution numerical simulations. We present the data structures and algorithms of our approach, focusing on the optimizations required to render them computationally efficient on CPUs and GPUs alike. We provide a scalable implementation in the OpenFPM software library for HPC. We benchmark our implementation on up to 16 Nvidia GTX 1080 GPUs and up to 64 Nvidia A100 GPUs showing state-of-the-art scalability (68% to 96% parallel efficiency) on three benchmark problems. On a single GPU, our implementation is 14 to 140-fold faster than on a multi-core CPU.
Cover Image
Affiliated With Sbalzarini
Selected By
Acknowledged Services Scientific Computing Facility
Publication Status Published
Edoc Link
Sfx Link
DOI 10.1007/978-3-030-78713-4_15
PubMed ID
WebOfScience Link
Alternative Full Text URL
Display Publisher Download Only true
Visible On MPI-CBG Website true
PDF Downloadable false
Created By sbalzari
Added Date 2021-06-17
Last Edited By thuem
Last Edited Date 2022-01-04 11:14:11.106
Library ID 8080
Document ID
Entry Complete true
eDoc Compliant true
Include in Edoc Report true
In Pure true
Ready for eDoc Export false
Author Affiliations Complete false
Project Name
Project URL
Grant ID
Funding Programme
Funding Organisation