Automatic mapping of Fortran programs onto clusters with graphics processing units
DOI:
https://doi.org/10.14529/cmse140305Keywords:
DVM for Heterogeneous systems, Fortran DVMH, hybrid computational systems with accelerators, GPU, Fortran, OpenMPAbstract
The paper shows results of exploitation of automated parallelization system SAPFOR for sequential programs parallelization for cluster with GPUs, including programs with regular data dependencies. The system translates Fortran program into Fortran DVMH program. Produced program can be executed on cluster. Fortran DVMH language, its compilers and debugging tools are bundled within DVM-system. Performed transformations with original programs are considered. Programs which use a variety of parallel programming technologies are emitted by the system. Source code characteristics are considered. Performance analysis is done on CPUs and GPUs of the K-100 computational cluster.
References
Bakhtin V.A., Borodich I.G., Kataev N.A, Klinov M.S., Kovaleva N.V., Krukov V.A., Podderyugina N.V. Dialog s programmistom v sisteme avtomatizacii rasparallelivania SAPFOR [Dialog with programmer in automated parallelization system SAPFOR]. Vestnik Nizhegorodskogo universiteta im. N.I. Lobachevskogo [Vestnik of Lobachevsky State University of Nizhni Novgorod]. Nizhni Novgorod, Nizhni Novgorod University Press, 2012. No. 5 (2). P. 242–245.
Bakhtin V.A., Klinov M.S., Krukov V.A., Podderyugina N.V., Pritula M.N., Sazanov Y.L. Rasshirenie DVM-modeli parallel’nogo programmirovaniya dlya klasterov s get- erogennymi uzlami [Extension of the DVM parallel programming model for clusters with heterogeneous nodes]. Vestnik Yuzhno-Ural’skogo gosudarstvennogo universiteta, seriya “Matematicheskoe modelirovanie i programmirovanie”. Chelyabinsk, Publishing of the South Ural State University, 2012. No. 18 (277), Issue 12. P. 82–92.
Bakhtin V.A., Borodich I.G., Kataev N.A., Klinov M.S., Krukov V.A., Podderyugina N.V., Pritula M.N., Sazanov Y.L. Rasparallelivanie s pomosch’yu DVM-sistemy nekotoryh prilozheniy gidrodinamiki dlya klasterov s graficheskimi processorami [Parallelization of some CFD applications for clusters with graphics processing units using DVM-system]. Nauchniy servis v seti Internet: poisk novyh resheniy: Trudy Mezhdunarodnoy superkomp’yuternoy konferencii (17–22 sentyabrya 2012, Novorossiysk) [Science service on Internet: search for new decisions: Proceedings of International supercomputer conference (17–22 of September, 2012, Novorossiysk)]. Moscow, Moscow Uni- versity Press, 2012. P. 444–450.
Niziev V.G., Koldoba A.V., Mirzade F.H., Panchenko V.Y., Poveschenko Y.A., Popov M.V. Chislennoe modelirovanie plavleniya dvuhkomponentnyh poroshkov pri lazernom spekanii [Numerical modeling of melting process of two-component powders in laser ag- glomeration]. Matematicheskoe modelirovanie [Mathematical modeling]. 2011. Vol. 23, No. 4. P. 90–102.
Aleksahin V.F., Bakhtin V.A., Zhukova O.F., Kolganov A.S., Krukov V.A., Podderyugi- na N.V., Pritula M.N., Savitskaya O.A., Shubert A.V. Rasparallelivanie na graficheskie processory testov NAS NPB 3.3.1 na yazyke Fortran DVMH [Parallelization of NAS NPB 3.3.1 benchmarks on graphics processing units using Fortran DVMH language]. Parallel’nye vychislitel’nye tehnologii (PaVT’2014): trudy mezhdunarodnoy nauchnoy konferencii (1–3 aprelya 2014, Rostov-na-Donu) [Parallel computing technologies (PaCT’2014): proceedings of international science conference (1–3 of April, 2014, Rostov- on-Don)]. Chelyabinsk, Publishing of the South Ural State University, 2014. P. 30–41.
Gibridniy vychislitel’niy klaster K-100 [Hybrid computational cluster K-100]. URL: http://www.kiam.ru/MVS/resourses/k100.html (accessed 04.08.2014).
