Claritas / Overview / Cluster Support

GLOBE Claritas Cluster Support

GLOBE Claritas provides various levels of support for Linux clusters. At the most general level, the user can include a block of processing modules within the MPISTART and MPIEND modules, and the processing on that block will be shared amongst the requested slave processors. One of the beauties of the MPISTART/END system is that the user doesn't need to modify the modules inside the MPI block in any way.

  • Support for MPI Linux Clusters
  • General support for most pre-stack processing modules
  • Minimal user setup required
  • OpenPBS interface provides sophisticated batch queuing
  • Torque interface provides sophisticated batch queuing

Full scale testing was undertaken on one of the largest commercially available supercomputers outside of the USA; the New Zealand Supercomputer Centre (NZSC). The NZSC has over 1600 Intel Xeon Blade processors and is a processing on demand facility.

Running a CPU-intensive sequence on 20 processors was 17 times faster than running on a single processor. Taking into account the time required to distribute the executable and data across the cluster, there was a linear speed up, relative to the number of processors used.

The second phase of the cluster-enabled development is the provision of a complete suite of computer-intensive applications for specific tasks such as pre- and post-stack 3D migration. The migrations are presently undergoing production scale testing.

GLOBE Claritas is now available on a processing on demand basis.