APPLU
Sisira Weeratunga
Computational Fluid Dynamics and Computational Physics
Solution of five coupled nonlinear PDE's, on a 3-dimensional logically structured grid, using an implicit psuedo-time marching scheme, based on two-factor approximate factorization of the sparse Jacobian matrix. This scheme is functionally equivalent to a nonlinear block SSOR iterative scheme with lexicographic ordering. Spatial discretization of the differential operators are based on second-order accurate finite volume scheme. Insists on the strict lexicographic ordering during the solution of the regular sparse lower and upper triangular matrices. As a result, the degree of exploitable parallelism during this phase is limited to O(N**2) as opposed to O(N**3) in other phases and it's spatial distribution is non-homogenous. This fact also creates challenges during the loop re-ordering to enhance the cache locality.
There are two approaches to control the memory and run-time requirements of the benchmark:
The program is capable of automatically verifying whether a given run conforms to the specification of the benchmark by using internally stored reference solutions. However, these reference solutions are available only for a fixed number of mesh size/time steps pairs. If the input data does not correspond to any of the internally stored reference solutions, the verification test is not performed. Otherwise, the output indicates whether or not the run was successfull in meeting the requirements of the verifications tests. To conform to the specification of the benchmark, a run should successfully pass all three verification tests. Failure in any one or more tests indicates non-conformance with the specifications.
Fortran 77
None
E. Barszcz, R. Fatoohi, V. Venkatkrishnan and S. Weeratunga
"Solution of Regular Sparse Triangular Systems on Vector
and Distributed-Memory Multiprocessors",
Rept. No: RNR-93-007, NASA Ames Research Center, 1993
http://www.nas.nasa.gov/Pubs/TechReports/ebarszcz/RNR-93-007/RNR-93-007.html