Computational Science Technical Note CSTN-125

CSTN Home

Asynchronous Communication for Finite-Difference Simulations on GPU Clusters using CUDA and MPI

D. P. Playne and K. A. Hawick

Archived January 2011

Abstract

Graphical processing Units (GPUs) are finding widespread use as accelerators in computer clusters. It is not yet trivial to program applications that use multiple GPU-enabled cluster nodes efficiently. A key aspect of this is managing effective communication between GPU memory on separate devices on separate nodes. We develop an algorithmic framework for Finite-Difference numerical simulations that would normally require highly synchronous data-parallelism so they can effectively use loosely coupled GPU-enabled cluster nodes. We employ asynchronous communications and appropriate memory overlay of computations and communications to hide latency.

Keywords: GPU; asynchronous communications; clusters; CUDA; MPI.

Full Document Text: PDF version.

Citation Information: BiBTeX database for CSTN Notes.

BiBTeX reference:

@INPROCEEDINGS{CSTN-125,
  author = {D. P. Playne and K. A. Hawick},
  title = {Asynchronous Communication for Finite-Difference Simulations on GPU
	Clusters using CUDA and MPI},
  booktitle = {Proc. International Conference on Parallel and Distributed Processing
	Techniques and Applications (PDPTA'11)},
  year = {2011},
  number = {PDP2793},
  pages = {169-174},
  address = {Las Vegas, USA},
  month = {18-21 July},
  publisher = {CSREA},
  timestamp = {2011.05.16}
}


[ CSTN Index | CSTN BiBTeX ]