Submitting Job
Execution Node
There are three machines networked together.
The AUTO mode selects the least busiest machine.
Regardless of which machine is selected,
Node-1 is where the final input/output files are stored.
When other nodes are selected, the execution results
will be automatically synchronized with Node-1 in real time.
Manual Sync
This option really should not be necessary under normal circumstances
because the files are frequently synchronized automatically.
However, it may be useful in case of power failures or system shutdowns
where a program terminates
prematurely without synchronization.
This option will not be shown when the job is run on Node-1.
Resources
The submitted job will start running almost immediately.
CPU cycles and memory are dynamically shared with all other running jobs.
The resource allocation can be monitored
from the job status panel. This panel also allows one to
cancel a running job. In order to protect
anonymity, the first four letters
of the jobname are masked in the job status display. To kill a job, one would
need to know the full name.
If the job did not run as expected:
- Check the "Console Output" for any clues.
- Make sure there are no blank lines or funny characters in the input boxes.
- Plot the structures and dispersion data to make sure they were
interpreted correctly.