Run WRF in the Cloud with Sabalcore HPC
Sabalcore hosts the popular Weather Research Forecast (WRF) model as part of our High Performance Computing services. Run any WRF core on-demand using our customizable and easy to use environment. We have the latest versions installed plus everything you need to build and run WRF in a production environment.
WRF at Sabalcore – Performance Matters
Three important aspects that a HPC cloud must have to run WRF at the best efficiency possible – Fast interconnect (network), fast file system, and WRF must be compiled and installed correctly to take full advantage of the HPC system. These three attributes are essential for achieving results.
Our WRF installations are finely tuned for optimum performance. We have dozens of pre-built packages ready to run including WPS, ARW, NMM, WRFDA, HWRF, 3D-var, and more. If you need a custom build, we offer to compile it for your, while the advanced WRF user has the option to compile and install their own.
Our hardware infrastructure is non-virtual. Why is this better? A non-virtual infrastructure provides the best performance possible because WRF gets full access to the CPU cores, network, and storage targets without interference from a hypervisor (virtual machine). All our compute nodes are bare-metal and have ultra low-latency Infiiniband interconnects for outstanding parallel processing performance which is essential for WRF.
WRF at Sabalcore
- WRF is pre-installed. Begin running in just a few minutes.
- Build your own WRF using your REGISTRY or have us do it for you.
- Support for fully automated execution of WRF jobs.
- Unrestricted job execution (no wall-time requirement)
Turn-key – everything is pre-installed, configured, and ready to go including the WRF model, the WRF Preprocessing System (WPS), and graphics softwares
Automation – ability to automate entire workload from downloading of Met data to pre and post processing
Reliability – our HPC cloud is designed for around-the-clock production work-loads
- Single origin Cloud – you know exactly where and who manages the cloud.
- “Bare-metal” performance – Our systems are non-virtual. There is no impeding hypervisor to affect performance.
- Infiniband connectivity between compute nodes for superior MPI performance
- Parallel File System – localized and direct-attached, provides outstanding data performance for WRF. This is essential for large production runs such as hind-casts.
- High-memory compute nodes – up to 8GB of non-virtual RAM per core provides plenty of memory for high-resolution models.
- Accelerated data transfer – download/upload your data to Sabalcore 2x faster using our proprietary client
We have supported WRF users from around the world for many years. Consider Sabalcore an extension of your business and a valued partner.
WRF can generate atmospheric simulations using real data (observations, analyses) or idealized conditions. WRF offers operational forecasting a flexible and computationally-efficient platform, while providing recent advances in physics, numerics, and data assimilation. WRF is designed to serve both atmospheric research and operational forecasting needs. The full WRF package is available including the WRF model, the WRF Preprocessing System (WPS), and graphics software are available at Sabalcore.