Accurate Predictions of Parallel Program Execution Time

1995 ◽  
Vol 25 (1) ◽  
pp. 16-30 ◽  
Author(s):  
M.A. Driscoll ◽  
W.R. Daasch
Keyword(s):  
Execution Time ◽  
Parallel Program ◽  
Program Execution

scholarly journals Optimization of modeling the water-oil mixture filtration in elastic porous media for clusters with Xeon Phi accelerators

2015 ◽  
pp. 177-186
Author(s):  
С.Е. Киреев
Keyword(s):  
Porous Media ◽  
Execution Time ◽  
Multiphase Flows ◽  
Optimization Techniques ◽  
Parallel Program ◽  
Programming Models ◽  
Program Execution ◽  
Xeon Phi ◽  
Intel Xeon Phi ◽  
Intel Xeon

На основе разработанного ранее программного комплекса для моделирования многофазных потоков в деформируемой пористой среде реализован новый параллельный программный комплекс, оптимизированный для запуска на кластере с ускорителями Intel Xeon Phi. Рассмотрены различные способы оптимизации, специфичные для данного ускорителя, и их влияние на время работы программы. Выполнено сравнение различных способов использования ускорителей в составе кластера: симметричного режима и режима offload. Получены оценки ускорения и эффективности при использовании различного числа узлов кластера. On the basis of a previously developed program for modeling the multiphase flows in finite-deformed porous media, a new parallel program optimized for clusters with Intel Xeon Phi accelerators is implemented. Several optimization techniques specific for such accelerators are considered and their effect on the program execution time is discussed. A comparison of the symmetric and offload programming models for the accelerators is performed. The parallelization speedup and efficiency are estimatedwhen using various numbers of cluster's nodes.

SYMBOLIC PERFORMANCE ESTIMATION OF SPECULATIVE PARALLEL PROGRAMS

2003 ◽  
Vol 13 (04) ◽  
pp. 513-524 ◽  
Author(s):  
H. GAUTAMA ◽  
A. J. C. VAN GEMUND
Keyword(s):  
Performance Prediction ◽  
Execution Time ◽  
Time Distribution ◽  
Prediction Accuracy ◽  
Analytical Approach ◽  
Parallel Programs ◽  
Performance Estimation ◽  
Program Execution ◽  
Empirical Distributions ◽  
Measurement Results

Speculative parallelism refers to searching in parallel for a solution, such as finding a pattern in a data base, where finding the first solution terminates the whole parallel process. Different performance prediction methods are required as compared to traditional parallelism. In this paper we introduce an analytical approach to predict the execution time distribution of data-dependent parallel programs that feature N-ary and binary speculative parallel compositions. The method is based on the use of statistical moments which allows program execution time distribution to be approximated at O(1) solution complexity. Measurement results for synthetic distributions indicate an accuracy that lies in the percent range while for empirical distributions on internet search engines the prediction accuracy is acceptable, provided sufficient workload unimodality.

Sign in / Sign up

Export Citation Format

Share Document