Department of Electrical and Computer Engineering
Information Infrastructure Institute

Documentation

CyBlue consists of 1024 compute nodes, supported by 64 I/O nodes. Each node contains two dual-core PPC440 CPUs running at 700Mhz, with 512MB of RAM per node.

The head node is an IBM eServer OpenPower 720s with dual dual-core 1.50GHz 64-bit POWER5 processors and 4GB of RAM, running the SuSE Linux Enterprise Server 9 operating system.

Remote access is only permitted via SSH. The hostname you will SSH to is the 'frontend' of the Blue Gene, named cyblue.ece.iastate.edu. You will login with your Iowa State Net ID and password. The file storage provided is not the same as your ISU file storage. You will initially need to use SCP to copy files to the Blue Gene.

Remote access should be limited to compiling of existing code and the submission of jobs. Please do not use the frontend as base for the creation of new code. New code should be written and debugged elsewhere, and tested on the smaller clusters at Iowa State. When you wish to port your existing code base to run on the Blue Gene, that is appropriate use for coding on the frontend.

Besides the home directory space given, there is also a large quantity of storage (11TB) for use by various research projects in the directories /gpfs01 and /gpfs02. You can store and run jobs from those directories. It is recommended that if you are working on a shared research code base, that you store the code in the /gpfs0[1,2]/projectname/ directories instead of your home directory. I/O speed to /gpfs01 and /gpfs02 is also much faster than your home directory.

Keep in mind no storage is backed up. Please copy your important research data, and job run output off-site to a secure location. It is your responsibility to make sure your data is backed up.

For detailed information on the Blue Gene, and how it works, see the IBM Redbooks.