KR23: How to partition the grid for parallel DPM Solver?
There are two options in DPM Parallel, Shared Memory and Message Passing.
1) Shared Memory: The Shared Memory option is based on a shared memory model. When using the Shared Memory option, the particle calculations are entirely managed by the FLUENT `host` process.
a. The Shared Memory option is suitable for computations where the machine running the FLUENT `host` process is an adequately large, shared memory, multiprocessor machine.
b. Make sure that the machine executing the host process has enough memory to accommodate the entire grid.
c. Scalability is an issue if machine running the FLUENT host process is not adequately large, shared memory, multiprocessor machine.
d. Shared Memory option cannot be used with any of the multiphase flow models (VOF, mixture, or Eulerian)
2) Message Passing: The Message Passing option enables cluster computing and also works on shared memory machines. With this option enabled, the compute `node` processes perform the particle work on their local partitions. Particle migration to other compute nodes is implemented using message passing primitives.
a. Do not partition the grid through injection location or location that will lead to large number of particle migration. Large number of particle migration will lead to more CPU time. Probability of particle loss during particle migration also increases with large number of particle migration
b. Avoid several parallel partitions meeting at edge or point, this will increase the probability of particle loss during migration
c. If there is large number of particles then particle calculation becomes time consuming. In this case it is advised to partition the grid such that almost equal number of particles is shared by each partition.
d. There are no special requirements for the host machine. The Message Passing option is turned on by default and is suitable for generic distributed memory cluster computing.
Never reorder the Grid when you are using DPM model
You may seamlessly switch between the Shared Memory option and the Message Passing option at any time during the FLUENT session.