Parallel Performance Measurement of Heterogeneous Parallel Systems with GPUs
Department of Computer and Information Science, University of Oregon, Eugene
International Conference on Parallel Processing (ICPP), 2011
@inproceedings{malony2011parallel,
title={Parallel Performance Measurement of Heterogeneous Parallel Systems with GPUs},
author={Malony, A.D. and Biersdorff, S. and Shende, S. and Jagode, H. and Tomov, S. and Juckeland, G. and Dietrich, R. and Poole, D. and Lamb, C.},
booktitle={Parallel Processing (ICPP), 2011 International Conference on},
pages={176–185},
year={2011},
organization={IEEE}
}
The power of GPUs is giving rise to heterogeneous parallel computing, with new demands on programming environments, runtime systems, and tools to deliver high-performing applications. This paper studies the problems associated with performance measurement of heterogeneous machines with GPUs. A heterogeneous computation model and alternative host-GPU measurement approaches are discussed to set the stage for reporting new capabilities for heterogeneous parallel performance measurement in three leading HPC tools: PAPI, Vampir, and the TAU Performance System. Our work leverages the new CUPTI tool support in NVIDIA’s CUDA device library. Heterogeneous benchmarks from the SHOC suite are used to demonstrate the measurement methods and tool support.
November 17, 2011 by hgpu
