DIRECTORY |
|
| DIRECTORY /my/path
|
| the directory in which the job should be executed
|
| This required keyword expects a string. Default value: .
|
INPUT_FILE_NAME |
|
| input_file my_input.inp
|
| the filename of the input file
|
| This required keyword expects a string. Default value: input.inp
|
JOB_ID |
|
| JOB_ID 13
|
| An ID used to indentify a job in DEPENDENCIES. JOB_IDs do not need to be unique, dependencies will be on all jobs with a given ID. If no JOB_ID is given, the index of the &JOB section in the input file will be used.
|
| This optional keyword expects an integer |
DEPENDENCIES |
|
| DEPENDENCIES 13 1 7
|
| specifies a list of JOB_IDs on which the current job depends. The current job will not be executed before all the dependencies have finished. The keyword requires a MASTER_SLAVE farming run. Beyond the default case, some special cases might arise: 1) circular dependencies will lead to a deadlock. 2) This keyword is not compatible with CYCLE. 3) MAX_JOBS_PER_GROUP is ignored (though only a total of MAX_JOBS_PER_GROUP*NGROUPS jobs will be executed) 4) dependencies on jobs that will not be executed (due to RESTART or MAX_JOBS_PER_GROUP) are ignored. Additionally, note that, on some file systems, output (restart) files might not be immediately available on all compute nodes,potentially resulting in unexpected failures.
|
| This optional keyword expects a list of integers |