1887

Abstract

Summary

A Roofline model is a tool that simultaneously considers the program that is being optimized and the underlying hardware architecture to be used for the execution. The roofline model provides the maximum performance that can be achieved for a given “Operational Intensity”, that is the ratio between floating point operations and bytes transferred from/to main memory. Moreover, it provides a valuable insight on your application that not only properly drives the optimization process, but gives the developer a good hint about when the code optimization is no longer a cost effective alternative.

In this work we show the process we have followed to optimize a production-ready, elastic anisotropic code, almost doubling its performance in a few series of steps and in ashort period of time. This was done by means of properly reading the roofline model, obtained by measuring the application behaviour. Our results are presented for the Xeon E5670 processor from Intel, but the same methodology could be applied for other HPC hardware architectures.

Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.201413288
2015-06-01
2020-03-30
Loading full text...

Full text loading...

References

  1. Araya-Polo, M., Rubio, F., de la Cruz, R., Hanzich, M., Cela, J.M. and Scarpazza, D.P.
    [2009] 3d seismic imaging through reverse-time migration on homogeneous and heterogeneous multi-core processors. Sci. Program., 17(1–2), 185–198, ISSN 1058-9244.
    [Google Scholar]
  2. Araya-Polo, M. et al.
    [2011] Assessing accelerator-based hpc reverse time migration. IEEE Transactions on Parallel and Distributed Systems, 22(1), 147–162, ISSN 1045-9219, doi:http://doi.ieeecomputersociety.org/10.1109/TPDS.2010.144.
    [Google Scholar]
  3. Ortiz, D. and Santiago, N.
    [2008] Impact of source code optimizations on power consumption of embedded systems. Circuits and Systems and TAISA Conference, 2008. NEWCAS-TAISA 2008. 2008 Joint 6th International IEEE Northeast Workshop on, 133–136, doi:10.1109/NEWCAS.2008.4606339.
    https://doi.org/10.1109/NEWCAS.2008.4606339 [Google Scholar]
  4. Rubio, F., Hanzich, M., Farrés, A., de la Puente, J. and Cela, J.M.
    [2014] Finite-difference staggered grids in GPUs for anisotropic elastic wave propagation simulation. Computers & Geosciences, 70(0), 181–189, ISSN 0098-3004, doi:http://dx.doi.org/10.1016/j.cageo.2014.06.003.
    [Google Scholar]
  5. Virieux, J. and Operto, S.
    [2009] An overview of full-waveform inversion in exploration geophysics. Geophysics, 74(6), WCC1–WCC26, doi:10.1190/1.3238367.
    https://doi.org/10.1190/1.3238367 [Google Scholar]
  6. Williams, S., Waterman, A. and Patterson, D.
    [2009] Roofline: An insightful visual performance model for multi-core architectures. Commun. ACM, 52(4), 65–76, ISSN 0001-0782, doi:10.1145/1498765.1498785.
    https://doi.org/10.1145/1498765.1498785 [Google Scholar]
http://instance.metastore.ingenta.com/content/papers/10.3997/2214-4609.201413288
Loading
/content/papers/10.3997/2214-4609.201413288
Loading

Data & Media loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error