View in EDS

A PARTITIONER-CENTRIC MODEL FOR STRUCTURED ADAPTIVE MESH REFINEMENT PARTITIONING TRADE-OFF OPTIMIZATION: PART I.

Saved in:
Bibliographic Details
Title: A PARTITIONER-CENTRIC MODEL FOR STRUCTURED ADAPTIVE MESH REFINEMENT PARTITIONING TRADE-OFF OPTIMIZATION: PART I.
Authors: Steensland, Johan1 JSTEENS@CA.SANDIA.GOV, Ray, Jaideep2
Source: International Journal of High Performance Computing Applications. Winter2005, Vol. 19 Issue 4, p409-422. 14p.
Subjects: Program transformation, Automation, Computer simulation, Computer programmers, High performance computing, Computer-aided design
Abstract: The article focuses on the development of a model that predicts the suitability of a structured adaptive mesh for purely domain-based or patch-based decomposition. The model could be used to determine one of the parameters used to configure a partitioner or choose the optimal one for a given problem. The predictions were validated against data obtained from four different SAMR simulations. Optimal partitioning of structured adaptive mesh applications necessitates dynamically determining and optimizing for the most time-inhibiting factor, such as load imbalance and communication volume. This model contributes to the meta-partitioner, able to select and configure the optimal partitioner based on the mesh configuration, the simulation and computer characteristics. The results show that the proposed model generally captures the inherent optimization-need in structured adaptive mesh refinement applications. The article concludes that the model could be useful for decreasing execution time for large SAMR applications.
Database: Engineering Source
Description
Abstract:The article focuses on the development of a model that predicts the suitability of a structured adaptive mesh for purely domain-based or patch-based decomposition. The model could be used to determine one of the parameters used to configure a partitioner or choose the optimal one for a given problem. The predictions were validated against data obtained from four different SAMR simulations. Optimal partitioning of structured adaptive mesh applications necessitates dynamically determining and optimizing for the most time-inhibiting factor, such as load imbalance and communication volume. This model contributes to the meta-partitioner, able to select and configure the optimal partitioner based on the mesh configuration, the simulation and computer characteristics. The results show that the proposed model generally captures the inherent optimization-need in structured adaptive mesh refinement applications. The article concludes that the model could be useful for decreasing execution time for large SAMR applications.
ISSN:10943420
DOI:10.1177/1094342005056140