SCATTER(6) | SCATTER(6) |
The options available for scatter:
The scatter command is designed with the view to be used with the SCore-D parallel job management facility (see also scrun(1) man page). scatter copies a file out of a cluster into cluster hosts in a scalable way. Or, users can run normal sequential programs or Unix commands run on a cluster in parallel using the system(6) SCore-D command.
Here are some examples:
$ scrun -nodes=8 scatter == a.out < my.datIn this example, the standard input file my.dat is fed as a standard input for all processes of a.out running on a cluster.
$ scrun -nodes=16x1 scatter -file cluster.dat :: a.out < local.datIn this case, the local file local.dat is copied to the file cluster.dat in a temporary directory under /var/scored on all 16 hosts, and then the user program a.out is executed on a cluster so that the a.out program can read the broadcasted file.