The National Geoelectromagnetic Facility provides cyberinfrastructure resources to users. This includes

a) Field computers and instrument control/data processing software - typically MacBook or MacBook Air notebooks with LabView Virtual Instruments for operating and/or setting up Narod NIMS MT systems and Zonge Zen Rx6 EM receivers. Field systems can also be equipped with MATLAB scripts that interface to the Oregon State University MT data processing software package.

b) High performance computation - with the collaboration of NGF personnel it is possible to carry out research programs that use our massively parallel computing systems, "parakepler" and "paracuda". As presently configured, up to 56 parallel CPU threads are supported on our Colfax International ( CX4850-2T-X6 hybrid CPU/GPU system, with two Intel Zeon E5-2697V3 2.6 GHz 14-core processors served by 768 GB of shared DDR4 2133 MHz RAM. Each processor thread has access to a farm of 8 NVidia Kepler K80 general purpose GPU's providing 30,000 parallel stream processors, on a high speed backplane. "Paracuda", our 2nd workgroup hybrid server, is pictured below. Paracuda is available for hybrid CPU/GPU code development purposes, and also to run codes requiring MS Windows 10, which is supported in a virtualization environment.

NGF's Colfax CXT-8000 hybrid CPU/GPU server with 5 Fermi-class and 3 Tesla-class GPU's installed.

The NGF supports parallel programming development using Portland Group's CUDA-enabled suite of Fortran compilers, as well as C and C++ languages, all with libraries and support packages for parallel coding. CUDA-enabled Fortran includes CUDA Fortran, a CUDA specific dialect with low-level GPU optimization capabilities, and a directive-based ANSI standard Fortran90/95 compiler with directives very similar to OpenMP. OpenMP and MPI are also supported for CPU parallel programming. Torque has also been installed to enable PBS cluster-based multiple core job sharing. For more information, contact the NGF directly.

The NGF thanks the US Department of Energy National Energy Technology Laboratory, and industrial partners Pacific Gas & Electric, SAIC, Zonge International, Inc., and NVidia Corporation for contributions that made it possible to build and upgrade these systems. The NGF acknowledges the support of the National Science Foundation for resources used to create instrument control interfaces and field data processing software development.