Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Skip to main content
Springer Nature Link
Log in

JAD-Based SpMV Kernels Using Multiple-Precision Libraries for GPUs

  • Conference paper
  • First Online:
Supercomputing (RuSCDays 2021)

Part of the book series: Communications in Computer and Information Science ((CCIS,volume 1510))

Included in the following conference series:

  • 782 Accesses

Abstract

Sparse matrix computations often arise in real-world applications, and sparse matrix-vector multiplication (SpMV) is one of the key operations in linear algebra. At the same time, single and double precision arithmetic, which is the most common in scientific computing and natively supported by hardware and programming languages, introduces round-off errors that may affect the SpMV results and in some cases cause serious problems. In this paper we implement SpMV kernels for graphics processing units using available software libraries that support multiple-precision computation. We use the Jagged Diagonal (JAD) sparse matrix storage format, which is very close to CSR in terms of memory consumption, but provides efficient accesses to nonzero matrix entries. We evaluate the implemented kernels on NVIDIA RTX 2080 for various sparse matrices from the SuiteSparse Matrix Collection.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution

Subscribe and save

Springer+ Basic
$34.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 84.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 109.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Similar content being viewed by others

Notes

  1. 1.

    https://github.com/skystar0227/CUMP.

  2. 2.

    https://github.com/kisupov/mpres-blas.

  3. 3.

    https://homepages.laas.fr/mmjoldes/campary.

References

  1. Ahmed, N., Mateev, N., Pingali, K., Stodghill, P.: A framework for sparse matrix code synthesis from high-level specifications. In: SC 2000: Proceedings of the 2000 ACM/IEEE Conference on Supercomputing, pp. 58–58 (2000). https://doi.org/10.1109/SC.2000.10033

  2. Bell, N., Garland, M.: Implementing sparse matrix-vector multiplication on throughput-oriented processors. In: Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis, Portland, OR, USA, pp. 1–11, November 2009. https://doi.org/10.1145/1654059.1654078

  3. Furuichi, M., May, D.A., Tackley, P.J.: Development of a stokes flow solver robust to large viscosity jumps using a Schur complement approach with mixed precision arithmetic. J. Comput. Phys. 230(24), 8835–8851 (2011). https://doi.org/10.1016/j.jcp.2011.09.007

    MathSciNet  MATH  Google Scholar 

  4. Hida, Y., Li, X.S., Bailey, D.H.: Algorithms for quad-double precision floating point arithmetic. In: Proceedings of the 15th IEEE Symposium on Computer Arithmetic, Vail, CO, USA, pp. 155–162, June 2001. https://doi.org/10.1109/ARITH.2001.930115

  5. Hishinuma, T., Hasegawa, H., Tanaka, T.: SIMD parallel sparse matrix-vector and transposed-matrix-vector multiplication in DD precision. In: Dutra, I., Camacho, R., Barbosa, J., Marques, O. (eds.) VECPAR 2016. LNCS, vol. 10150, pp. 21–34. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-61982-8_4

    Google Scholar 

  6. Isupov, K., Knyazkov, V.: Multiple-precision BLAS library for graphics processing units. In: Voevodin, V., Sobolev, S. (eds.) RuSCDays 2020. CCIS, vol. 1331, pp. 37–49. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-64616-5_4

    Google Scholar 

  7. Joldes, M., Muller, J.-M., Popescu, V., Tucker, W.: CAMPARY: Cuda multiple precision arithmetic library and applications. In: Greuel, G.-M., Koch, T., Paule, P., Sommese, A. (eds.) ICMS 2016. LNCS, vol. 9725, pp. 232–240. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-42432-3_29

    Google Scholar 

  8. Li, R., Saad, Y.: GPU-accelerated preconditioned iterative linear solvers. J. Supercomput. 63(2), 443–466 (2013). https://doi.org/10.1007/s11227-012-0825-3

    Google Scholar 

  9. Lu, M., He, B., Luo, Q.: Supporting extended precision on graphics processors. In: Sixth International Workshop on Data Management on New Hardware (DaMoN 2010), Indianapolis, Indiana, USA, pp. 19–26, June 2010. https://doi.org/10.1145/1869389.1869392

  10. Masui, K., Ogino, M., Liu, L.: Multiple-precision iterative methods for solving complex symmetric electromagnetic systems. In: van Brummelen, H., Corsini, A., Perotto, S., Rozza, G. (eds.) Numerical Methods for Flows. LNCSE, vol. 132, pp. 321–329. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-30705-9_28

    Google Scholar 

  11. Nakayama, T., Takahashi, D.: Implementation of multiple-precision floating-point arithmetic library for GPU computing. In: Proceedings of the 23rd IASTED International Conference on Parallel and Distributed Computing and Systems (PDCS 2011), Dallas, USA, pp. 343–349, December 2011. https://doi.org/10.2316/P.2011.757-041

  12. Omondi, A., Premkumar, B.: Residue Number Systems: Theory and Implementation. Imperial College Press, London (2007)

    MATH  Google Scholar 

  13. Saad, Y.: Iterative Methods for Sparse Linear Systems, 2nd edn. SIAM, Philadelphia (2003)

    MATH  Google Scholar 

  14. Saito, T., Ishiwata, E., Hasegawa, H.: Analysis of the GCR method with mixed precision arithmetic using QuPAT. J. Comput. Sci. 3(3), 87–91 (2012)

    Google Scholar 

Download references

Acknowledgements

This work was funded by the Russian Science Foundation grant number 20-71-00046.

Author information

Authors and Affiliations

  1. Vyatka State University, Kirov, Russia

    Konstantin Isupov & Ivan Babeshko

Authors
  1. Konstantin Isupov
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ivan Babeshko
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Konstantin Isupov .

Editor information

Editors and Affiliations

  1. RCC MSU, Moscow, Russia

    Vladimir Voevodin

  2. RCC MSU, Moscow, Russia

    Sergey Sobolev

Rights and permissions

Reprints and permissions

Copyright information

© 2021 Springer Nature Switzerland AG

About this paper

Check for updates. Verify currency and authenticity via CrossMark

Cite this paper

Isupov, K., Babeshko, I. (2021). JAD-Based SpMV Kernels Using Multiple-Precision Libraries for GPUs. In: Voevodin, V., Sobolev, S. (eds) Supercomputing. RuSCDays 2021. Communications in Computer and Information Science, vol 1510. Springer, Cham. https://doi.org/10.1007/978-3-030-92864-3_12

Download citation

  • DOI: https://doi.org/10.1007/978-3-030-92864-3_12

  • Published:

  • Publisher Name: Springer, Cham

  • Print ISBN: 978-3-030-92863-6

  • Online ISBN: 978-3-030-92864-3

  • eBook Packages: Computer ScienceComputer Science (R0)

Publish with us

Policies and ethics

Search

Navigation