UMT performs the solution of time-dependent, energy-dependent, discrete ordinates, and nonlinear radiation problems on three- dimensional unstructured spatial grids on large distributed-memory multi-node parallel computer systems with multiple cores per node. To achieve extreme scalability, the application exploits both spatial decomposition using message passing between nodes and a threading algorithm in angle within the node.
The project addresses the need to obtain accurate transport solutions to three- dimensional radiation transfer problems (i.e. the transfer of thermal photons). The solution calls for the use of an unstructured grid. This class of problems is characterized by tens of thousands of unknowns per zone and upwards of millions of zones, thus requiring large, scalable, parallel computing platforms. The package uses a combination of message passing and threading, utilizes the large distributed memory of the platform and provides unprecedented (weak) scaling to very large core counts.
Required Problem Sets
For UMT, small, large and grand-challenge problems have been defined. See the README.APEX file in the source distribution for details. The large definition will be used as the reference problem and the grand challenge definition is the target problem in the calculation of SSI.
Source and problem definitions can be downloaded here.
How to Build, Run and Verify
Refer to the README.APEX file in the source distribution.
UMT was developed by Lawrence Livermore National Laboratory.
|2/18/2016||Updated source to address missing .F90 files
Updated README.APEX file
Updated benchmarks directory to include example small problem
|1/28/2016||Source distribution provided|