Method of parallelization of loops for grid calculation problems on GPU accelerators

А.Yu. Doroshenko, O.G. Beketov

Abstract


The formal parallelizing transformation of a nest of calculation loop for SIMD architecture devices, particularly for graphics processing units applying CUDA technology and heterogeneous clusters is developed. Procedure of transition from sequential to parallel algorithm is described and illustrated. Serialization of data is applied to optimize processing of large volumes of data. The advantage of the suggested method is its applicability for transformation of data which volumes exceed the memory of operating device. The experiment is conducted to demonstrate feasibility of the proposed approach. Technique presented in the provides the basis for further practical implementation of the automated system for parallelizing of nested loops.

Problems in programming 2017; 1: 59-66


Keywords


automatic parallelization; loop optimization; general-purpose computing on graphics processing units

Full Text:

PDF (Ukrainian)

References


Doroshenko, A.Yu. & Shevchenko R.S. (2005) Symbolic computation system for dynamical application programming. Problems in programming. (4). P. 718–727. (in Russian)

Yatsenko, O.A. (2013) Integration of Software Tools of Algebra of Algorithms and Rewriting Terms for Development of Effective Parallel programs. Problems in programming. (2). P. 62–70. (in Russian).

Doroshenko, A.Yu., Beketov, O.G , Prusov, V.A., Tyrchak, Yu.M. & Yatsenko, O.A. (2014) Formalized design and generation of parallel programs for numerical weather forecast. Problems in programming. (2-3). P. 72–81. (in Ukrainian).

Doroshenko, А.Yu., Beketov, O.G., Ivaniv, R.B., Iovchev, V.O., Mironenko, I.O. & Yatsenko, O.A. (2015) Automated generation of parallel programs for graphics processing units based on algorithm schemes. Problems in programming. (1). P. 19–28. (in Ukrainian).

CUDA [Online] – Available from: http://www.nvidia.com/object/cuda_home_new.html

TESLA [Online] – Available from: http://www.nvidia.com/object/teslaservers.html

PIPS: Automatic Parallelizer and Code Transformation Framework [Online] – Available from: http://pips4u.org/

Prusov V.A. & Doroshenko A.Yu. (2006) Simulation of natural and anthropogenic processes in the atmosphere. Kyiv: Naukova Dumka. (in Ukrainian).

Prusov, V.A. & Snizhko, S.I. (2005) Mathematical modeling of atmospheric processes. Kyiv: NikaTsentr. (in Ukrainian).


Refbacks

  • There are currently no refbacks.