memory parallelism on shared memory parallel hardware (e.g., a desktop or workstation) or distributed
memory parallel hardware (e.g., a cluster). MDOF indicat es million degrees of freedom.
Table 5.1: Shared Memory Solver Selection Guidelines
Disk
(I/O)
Use
Ideal Model Memory Use
Size
Solver Typical Applications
Op-
tim-
Optimal
out-of-
100,000 DOF
to 5 MDOF
When robustness and solution
speed are required (nonlinear ana-
Sparse Dir-
ect Solver
core: 1 al
GB/MDOF
(works well
outside this
range)
lysis); for linear analysis where iterat-
ive solvers are slow to converge
(especially for ill-conditioned
(direct elim-
ination) out-
of-
In-core:
10
GB/MDOF
matrices, such as poorly shaped
elements).
core:
10
GB/MDOF
In-
core:
1
GB/MDOF
0.5
GB/MDOF
0.3 GB/MDOF
w/MSAVE,ON;
500,000 DOF
to 20
MDOF+
Reduces disk I/O requirement relat-
ive to sparse solver. Best for large
models with solid elements and fine
PCG Solver
(iterative
solver) 1 GB/MDOF
meshes. Most robust iterative solver
in ANSYS.
without
MSAVE
0.5
GB/MDOF
500,000 DOF 0.5 GB/MDOF
to 20
MDOF+
Best for single field problems -
(thermal, magnetics, acoustics, and
multiphysics). Uses a fast but simple
JCG Solver
(iterative
solver)
preconditioner with minimal
memory requirement. Not as robust
as PCG solver.
0.5
GB/MDOF
50,000 to 1.5 GB/MDOF
1,000,000+
DOF
More sophisticat ed preconditioner
than JCG. Best for more difficult
problems where JCG fails, such as
unsymmetric thermal analyses.
ICCG Solver
(iterative
solver)
0.5
GB/MDOF
50,000 to 1.5 GB/MDOF
1,000,000+
DOF
QMR Solver High-frequency electromagnetics.
(iterative
solver)
Table 5.2: Distributed Memory Solver Selection Guidelines
Disk
(I/O)
Use
Ideal Model Memory Use
Size
Solver Typical Applications
Op-
tim-
Optimal
out-of-
500,000 DOF
to 10 MDOF
Same as sparse solver but can also
be run on distributed memory par-
allel hardware systems.
Distributed
Memory
Sparse Dir-
ect Solver
al
out-
core: 1.5
GB/MDOF
(works well
outside this
range) on mas- of-
ter ma- core:
Release 15.0 - © SAS IP, Inc. All rights reserved. - Contains proprietary and confidential information
Solution