ORNL operates the ORNL Leadership Class Computing Facility (OLCF), one of DOE’s Advanced Scientific Computing Research (ASCR) facilities (https://science.energy.gov/ascr/facilities). OLCF is the home of the world’s past and present fastest super computers. Although these machines are not often required for routine calculations, the expertise of teams who are successfully awarded machine allocations will be leveraged in developing software and performing calculations on these leadership class machines.
Commodity Clusters for Production Calculations
ORNL has a large inventory of computer hardware that is actively managed. A subset of hardware accessible to projects with appropriate export control and IP environments for production-quality computational analysis is provided to ONRAMP users with over 13K CPU cores and 50 TB RAM.
Dedicated computer clusters can be procured and maintained at ORNL for projects so that work can proceed at the highest priority and sufficient computing capacity will be available for design iterations.
Secure Cloud Computing
Access to industry-class computers using a secure access protocol is available to certain foreign nationals who otherwise would not be provided access to certain software while ensuring that the requirements of Federal nuclear export control are satisfied. This provision is through RSICC and its International Secure Platform for Export-Controlled Computing Tools (InSPECT) capability and application-specific cloud license. The cloud license allows users to submit analysis to ORNL’s industry-class computing resource without having access to the application (SCALE, MCNP®, RELAP-5, etc.) executables or source code. This provides approved users with the fastest access to configuration- and export-controlled tools and analysis capabilities.
CADES
ORNL’s Compute and Data Environment for Science (CADES) provides eligible customers with an OpenStack cloud computing solution with customizable virtual machines (VMs). This resource, called CADES Cloud, enables customers to leverage self-service portals to rapidly request these VMs for production, testing, and development.