Getting Started Using Dawson

Once your user account is approved (see Acquiring An Account) you will be contacted by the CSCAMM SP/2 Systems Administrator to arrange creation and activation of a login account. Typically this just involves stopping by our offices to enter your password.

The newly created login account grants you a access the Front End Workstation to the CSCAMM IBM SP/2, this is the only machine able to submit jobs to run on the cluster.

Connecting

Hostname: dawson.cscamm.umd.edu
Login Protocols:  ssh, telnet, ftp

Login Account Information

All user accounts are set up with:

/usr/bin/tcsh as your shell

/gpfs/home/username as your home directory
(100 Megabyte Disk Quota)
Currently backed up weekly

/gpfs/scratch/username as your scratch directory
(no quota, but files automatically deleted)
Not backed up

/gpfs/projects/projectname/username for each project you are member
(quota shared with project)
Currently backed up weekly

Common applications are accessed through the following directories and should be part of your path:

Common documentation (manpage format) are available in the following directories and should be part of your MANPATH environment variable.

Installed Development Tools

The following IBM software development tools are installed

IBM C Compiler for AIX (cc,c89,xlc)
IBM VisualAge C++ Compiler for AIX (xlC)
IBM XL Fortran Compiler for AIX (f77,xlf,xlf90,xlf95)
IBM High Performance Fortran Compiler for AIX
IBM Parallel Operating Environment/MPI (mpcc,mpxlf,mpxlf90,mpxlf95)

Plus threading (AIX standard, and POSIX Draft 7 versions) options (_r,_r7).

GNU compilers are currently not being made available because they are inherently inferior to the above products for use on this system. If this causes difficulty contact the systems administrator.

Job Submission System

In order to ensure fair and efficient use of the cluster jobs are executed by submitting a job command file to a central scheduler which then queues and starts each job as conditions permit. The scheduler in use on dawson is the standard IBM LoadLeveler. It uses a backfill algorithm which runs shorter jobs ahead of  longer jobs if the available nodes would otherwise sit idle. Otherwise scheduling of jobs is first come, first served. We have defined six job queues to serve this cluster (see Using Queues on Dawson).

How resource allocations works

Every project is assigned an allocation of CPU hours per month that can be used. Every job submitted to LoadLeveler is tagged with an allocation account to track usage. If a project overruns it's allocation for a month jobs for that project will be rejected. If all of a project's allocation for a month is not used then the remainder is lost and will NOT carry over into the following month.

How to submit batch jobs

See Submitting Batch Jobs on Dawson.

How to submit interactive jobs

See Submitting Interactive Jobs on Dawson.

Documentation