Raghavan, Hari K and Vadhiyar, Sathish S (2015) Adaptive executions of hyperbolic block-structured AMR applications on GPU systems. In: INTERNATIONAL JOURNAL OF HIGH PERFORMANCE COMPUTING APPLICATIONS, 29 (2). pp. 135-153.
Full text not available from this repository. (Request a copy)Abstract
A block-structured adaptive mesh refinement (AMR) technique has been used to obtain numerical solutions for many scientific applications. Some block-structured AMR approaches have focused on forming patches of non-uniform sizes where the size of a patch can be tuned to the geometry of a region of interest. In this paper, we develop strategies for adaptive execution of block-structured AMR applications on GPUs, for hyperbolic directionally split solvers. While effective hybrid execution strategies exist for applications with uniform patches, our work considers efficient execution of non-uniform patches with different workloads. Our techniques include bin-packing work units to load balance GPU computations, adaptive asynchronism between CPU and GPU executions using a knapsack formulation, and scheduling communications for multi-GPU executions. Our experiments with synthetic and real data, for single-GPU and multi-GPU executions, on Tesla S1070 and Fermi C2070 clusters, show that our strategies result in up to a 3.23 speedup in performance over existing strategies.
| Item Type: | Journal Article |
|---|---|
| Publication: | INTERNATIONAL JOURNAL OF HIGH PERFORMANCE COMPUTING APPLICATIONS |
| Publisher: | SAGE PUBLICATIONS LTD |
| Additional Information: | Copy right for this article belongs to the SAGE PUBLICATIONS LTD, 1 OLIVERS YARD, 55 CITY ROAD, LONDON EC1Y 1SP, ENGLAND |
| Keywords: | Adaptive mesh refinement; GPU executions; dynamic load balancing; asynchronous executions of CPUs and GPUs; coalesced access |
| Department/Centre: | Division of Interdisciplinary Sciences > Supercomputer Education & Research Centre |
| Date Deposited: | 22 May 2015 07:09 |
| Last Modified: | 22 May 2015 07:09 |
| URI: | http://eprints.iisc.ac.in/id/eprint/51546 |
Actions (login required)
 |
View Item |
