Abaqus should be run using the following command:

abaqus job=<job> cpus=$RESCALE_CORES_PER_SLOT mp_mode=mpi interactive

This will run Abaqus on all cores specified in the hardware settings for the job. Note that is the filename of your .inp input file (.inp - without the extension).

You can also set the scratch directory by appending the command: scratch = path/to/file. For example:

abaqus job=s4b cpus=$RESCALE_CORES_PER_SLOT mp_mode=mpi scratch=$PWD/tmp interactive

Some common (optional) additional commands include: [gpus=number-of-gpgpus], [memory=memory-size], [output_precision={single | full}], [resultsformat={odb | sim | both}]. You can also SSH into your cluster and run abaqus -help to get the complete list of commands.

To execute a global + local (2 stage) model job, you must upload both the global and local .inp files as input files. Run the global and local models using the general command illustrated below:

Global Model: abaqus job=<global> interactive

Local Model: abaqus job=<local> globalmodel=<global> interactive

For example, if your global and local model is named input_global.inp and input_local.inp respectively, and you need it run in parallel with full output precision and double precision, the command would be:

#Global Model 
abaqus job=input_global cpus=$RESCALE_CORES_PER_SLOT mp_mode=mpi output_precision=full double=both parallel=domain interactive
#Local Model 
abaqus job=input_local cpus=$RESCALE_CORES_PER_SLOT mp_mode=mpi output_precision=full double=both parallel=domain globalmodel=input_global interactive

Rescale will run the commands sequentially, executing the global model followed by the local model.