You can run parallel fluent jobs on the cluster, by specifying the number of cores through the -n option of bsub. Fluent supports distributed memory parallel mode, which allows you to run single simulations across multiple compute nodes. Using fluent, you can choose between PCMPI and Intel MPI. On our HPC clusters, we noticed a problem with the CPU affinity settings, when using PCMPI. In some cases, all MPI processes were bound to a single processor core, which lead to a slow down for those jobs (by a factor that is equal to the number of cores, that users specified for the simulation). Therefore, we strongly recommend to not change any MPI settings through command line parameters. For parallel fluent jobs, the batch system automatically sets the MPI options such that Intel MPI is used.