6.7) ARC2 Architecture And SGE Parallel Environments (cont'd)

In addition to the standard Grid Engine submission syntax, outlined above, ARC2 also makes use of an alternative "nodes" syntax, which was developed and implemented at Leeds. This is designed to give jobs dedicated access to entire nodes. This should provide more predicable job performance, for instance due to placement and dedicated use of InfiniBand cards as well as providing a more flexible specification of processes or threads for hybrid OpenMP and MPI programming.

It can take either of the following forms:

• -l nodes=<w>[,ppn=<y>][,tpp=<z>]
• -l np=<x>[,ppn=<y>][,tpp=<z>]

Where:

• w = number of nodes requested
• x = number of processes requested
• y = number of processes per node
• z = number of threads per process

If y and z are omitted, Grid Engine sets y = number of cores in each machine, z = 1.
If y is present and z omitted, Grid Engine sets z = int(num cores / y).
If z is present and y omitted, Grid Engine sets y = int(num cores / z).
If using this syntax, the amount of memory available to the job is the total memory available on the node, i.e. 32GB.

These options are tabulated on the following page.