scholarly journals Dynamic Load Balancing Techniques for Particulate Flow Simulations

Computation ◽  
2019 ◽  
Vol 7 (1) ◽  
pp. 9 ◽  
Author(s):  
Christoph Rettinger ◽  
Ulrich Rüde
Keyword(s):  
Load Balancing ◽  
Graph Partitioning ◽  
Load Distribution ◽  
Dynamic Load Balancing ◽  
Second Step ◽  
Particulate Flows ◽  
Space Filling ◽  
Flow Simulations ◽  
Space Filling Curves ◽  
Multiphysics Simulations

Parallel multiphysics simulations often suffer from load imbalances originating from the applied coupling of algorithms with spatially and temporally varying workloads. It is, thus, desirable to minimize these imbalances to reduce the time to solution and to better utilize the available hardware resources. Taking particulate flows as an illustrating example application, we present and evaluate load balancing techniques that tackle this challenging task. This involves a load estimation step in which the currently generated workload is predicted. We describe in detail how such a workload estimator can be developed. In a second step, load distribution strategies like space-filling curves or graph partitioning are applied to dynamically distribute the load among the available processes. To compare and analyze their performance, we employ these techniques to a benchmark scenario and observe a reduction of the load imbalances by almost a factor of four. This results in a decrease of the overall runtime by 14% for space-filling curves.

Dynamic load balancing for direct-coupled multiphysics simulations

2020 ◽  
Vol 199 ◽  
pp. 104437 ◽  
Author(s):  
Ansgar Niemöller ◽  
Michael Schlottke-Lakemper ◽  
Matthias Meinke ◽  
Wolfgang Schröder
Keyword(s):  
Load Balancing ◽  
Dynamic Load ◽  
Dynamic Load Balancing ◽  
Multiphysics Simulations

RANDOM SEEKING: A GENERAL, EFFICIENT, AND INFORMED RANDOMIZED SCHEME FOR DYNAMIC LOAD BALANCING

2000 ◽  
Vol 11 (02) ◽  
pp. 231-246 ◽  
Author(s):  
NIHAR R. MAHAPATRA ◽  
SHANTANU DUTT
Keyword(s):  
Artificial Intelligence ◽  
Load Balancing ◽  
Branch And Bound ◽  
Operations Research ◽  
Dynamic Load ◽  
Load Distribution ◽  
Search Algorithms ◽  
Dynamic Load Balancing ◽  
Divide And Conquer ◽  
Branch And Bound Algorithms

We propose a completely general, informed randomized dynamic load balancing method called random seeking (RS) suitable for parallel algorithms with characteristics found in many search algorithms used in artificial intelligence and operations research and many divide-and-conquer algorithms. In it, source processors randomly seek out sink processors for load balancing by flinging "probe" messages. These probes not only locate sinks, but also collect load distribution information which is used to efficiently regulate load balancing activities. We empirically compare RS with a well-known randomized dynamic load balancing method, the random communication (RC) strategy, by using them in parallel best-first branch-and-bound algorithms on up to 512 processors of a parallel system. We find that the RC execution times are more than those of RS by 16–67% when used to perform combined dynamic quantitative and qualitative load balancing, and by 9–74% when used to perform only dynamic quantitative load balancing.

scholarly journals Applying graph partitioning methods in measurement-based dynamic load balancing

2011 ◽  
Author(s):  
Abhinav Bhatele ◽  
Sebastien Fourestier ◽  
Harshitha Menon ◽  
Laxmikant V. Kale ◽  
Francois Pellegrini
Keyword(s):  
Load Balancing ◽  
Dynamic Load ◽  
Graph Partitioning ◽  
Dynamic Load Balancing ◽  
Partitioning Methods

scholarly journals Load balancing using Hilbert space-filling curves for parallel shallow water simulations

2019 ◽  
pp. 75-87
Author(s):  
А.В. Чаплыгин ◽  
Н.А. Дианский ◽  
А.В. Гусев
Keyword(s):  
Hilbert Space ◽  
Load Balancing ◽  
Shallow Water ◽  
Shallow Water Equations ◽  
General Circulation ◽  
Sigma Model ◽  
Ocean Model ◽  
Computational Domain ◽  
Space Filling ◽  
Space Filling Curves

Представлен метод балансировки нагрузки вычислений с использованием кривых Гильберта применительно к параллельному алгоритму решения уравнений мелкой воды. Рассматриваемая система уравнений мелкой воды возникает в сигма-модели общей циркуляции океана INMOM (Institute of Numerical Mathematics Ocean Model) при разрешении гравитационных волн и является одним из основных блоков модели. Из-за наличия в океанах островов и берегов балансировка нагрузки вычислений на процессоры является особенно актуальной задачей. В качестве одного из таких методов был выбран метод балансировки нагрузки вычислений с использованием кривых Гильберта. Продемонстрирована большая эффективность этого метода по сравнению с равномерным разбиением без балансировки нагрузки и показано, что этот метод служит хорошей альтернативой библиотеке разбиений METIS. Оптимальность реализованного разбиения для мелкой воды точно соответствует оптимальности и для трехмерной сигма-модели INMOM в силу одинакового количества вертикальных уровней во всей расчетной области. This paper presents a method of load balancing using Hilbert space-filling curves applied to a parallel algorithm for solving shallow water equations. We consider the system of shallow water equations in the form presented in the ocean general circulation sigma-model INMOM (Institute of Numerical Mathematics Ocean Model). This system of equations is one of the basic blocks of the model. Due to land points in the computational grid, the load balancing is an especially urgent task. The method of load balancing using Hilbert space-filling curves is chosen as one of such methods. The paper demonstrates the greater efficiency of this method in comparison with the uniform partitioning without load balancing. It is shown that this method is a good alternative to the METIS standard library. Moreover, the optimality of the implemented partition for the shallow water equations exactly corresponds to the optimality for the INMOM three-dimensional sigma-model due to the same number of vertical levels in the entire computational domain.

Sign in / Sign up

Export Citation Format

Share Document