The distribution of computational load over different processing elements represents an important issue in parallel computing [1]. This is particularly relevant in thecase of parallel execution of structured grid computational models, such as CellularAutomata (CA) [2], where the domain space is partitioned in region assigned to theparallel computing nodes. Load balancing techniques are particularly effective whenstructured grid computational models are exploited to simulate topologically connectedphysical phenomena like lava or debris flows (e.g., [3]), in which the evolution develops in a usually small sub-region of the domain. In this work, we present a dynamicload balancing technique that can provide performance improvements in structuredgrid model development on distributed memory architectures by adopting the MPItechnology. First tests have demonstrated the usefulness of the feature in appreciablyreducing execution times in comparison with not-balanced parallel versions.
A Dynamic Load Balancing technique for Parallel Execution of Structured Grid Models
Giordano Andrea;
2019
Abstract
The distribution of computational load over different processing elements represents an important issue in parallel computing [1]. This is particularly relevant in thecase of parallel execution of structured grid computational models, such as CellularAutomata (CA) [2], where the domain space is partitioned in region assigned to theparallel computing nodes. Load balancing techniques are particularly effective whenstructured grid computational models are exploited to simulate topologically connectedphysical phenomena like lava or debris flows (e.g., [3]), in which the evolution develops in a usually small sub-region of the domain. In this work, we present a dynamicload balancing technique that can provide performance improvements in structuredgrid model development on distributed memory architectures by adopting the MPItechnology. First tests have demonstrated the usefulness of the feature in appreciablyreducing execution times in comparison with not-balanced parallel versions.File | Dimensione | Formato | |
---|---|---|---|
Abstract_LB_NUMTA2019.pdf
non disponibili
Licenza:
Nessuna licenza dichiarata (non attribuibile a prodotti successivi al 2023)
Dimensione
111.39 kB
Formato
Adobe PDF
|
111.39 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.