Published December 2, 2012
| Version v1
Working paper
Open
Scalable and Improved SuperLU on GPU for Heterogeneous Systems
Creators
- 1. Istanbul Technical University, National Center for High Performance Computing of Turkey (UHeM), Istanbul 34469, Turkey; Istanbul Technical University, Informatics Institute, Istanbul 34469, Turkey
Contributors
Others:
- 1. Istanbul Technical University, National Center for High Performance Computing of Turkey (UHeM), Istanbul 34469, Turkey; Istanbul Technical University,Department of Mathematics, Istanbul 34469, Turkey
- 2. Istanbul Technical University, National Center for High Performance Computing of Turkey (UHeM), Istanbul 34469, Turkey; Istanbul Technical University, Informatics Institute, Istanbul 34469, Turkey
Description
We consider a fast, robust and scalable solver using graphic processing units (GPU) as accelerators for a sparse linear system AX=B. In this project, a new parallel hybrid direct solver is designed and implemented on GPU for large sparse linear systems. In particular, we work on GPU programming using directive based Open ACC in order to obtain a scalable and improved SuperLU on CPU+GPU heterogeneous systems.
Project ID: FP7-INFRASTRUCTURES-2011-2, PRACE-2IP —PRACE - Second Implementation Phase Project
Files
ScalableSuperLUonGPU.pdf
Files
(187.8 kB)
Name | Size | Download all |
---|---|---|
md5:6c0bafeb6a55e44e445208225b0946fa
|
187.8 kB | Preview Download |