PARDISO is not supported in distributed mode. The Parallel Direct Sparse Solver (PARDISO) for Clusters from the Intel® oneMKL (Intel® oneAPI Math Kernel Library) is used instead if the Parallel Direct Sparse Solver for Clusters checkbox is selected; otherwise, MUMPS is used instead.

COMSOL Multiphysics can also run parametric sweeps using the distributed mode. This setting is available for all parametric solvers when Advanced Study Options is active. The simplest way to start a distributed parametric sweep is to select the Distribute parametric sweep checkbox in the Settings window for Cluster Computing in the Cluster Settings section. The simplest way to modify an existing model is to add the Cluster Computing study and select Compute in the study node’s Settings window. When running a parametric sweep in distributed mode, the memory usage is still limited by the memory size of a single computer node. Because the problems are solved in parallel, you will solve for all parameter values faster compared to solving on a single computer node.

If you want to combine the benefits from distributed parametric sweeps and the distributed solvers, it is possible to run in a hybrid mode where you can limit the maximum number of groups to distribute the sweep over. Use the Show Default Solver command to access this setting, which you find in the Cluster Settings section of the Settings window for the Parametric solver node.