|
The data structure of the mesh strongly influences the overall
performance of the engineering simulations as it is an infrastructure
system executing underneath to provide all the needed mesh-based
operations including distributed mesh operations on parallel computers
and dynamic mesh load balancing. Development of scalable applications
with complex distributed dynamic data structures is a challenging
undertaking due to complexity from non-trivial computational
capabilities.
|
|
As for the effort toward the scalable
engineering simulations on distributed environments, we addressed this
challenge by developing a distributed mesh data management
infrastructure that satisfies the needs of distributed domain of
applications. To achieve a good compromise between the storage and
computational costs, the distributed mesh data structure is designed to
be flexible in terms of mesh representation to easily switch between
various representations for different phases of an application. The FMDB
is embedded in the SCOREC simulation packages efficiently supporting
parallel automated adaptive analysis of meshes with billions of elements on massively parallel super computers such as ANL & CCNI BlueGene. |
|
FMDB is being used in various scientific/engineering applications as for a mesh data structure running underneath effectively supporting parallel mesh adaptive loop.
|
|
The FMDB is supported by the U.S. Department of Energy's
Scientific Discovery through Advanced Computing (SciDAC) program as part of the
Interoperable Tools for Advanced Petascale Simulations (ITAPS) center. The FMDB was developed to be
ITAPS compliant and represent core functionalities of the ITAPS meshing tools. |
|
|
* Example: a torus with four circular holes. The left is the initial
mesh and the right is an adapted mesh with two spherical shocks.
Different colors represent different processors.
|
