How is Gilbreth different than other Community Clusters?
Gilbreth differs from the previous Community Clusters in many significant aspects:
- Each Gilbreth compute nodes are equipped with a variety of Nvidia Tesla GPU accelerator cards which can significantly improve performance of compute-intensive workloads.
- Each Gilbreth front-end contains one Nvidia Tesla A30 accelerator card. This makes GPU code development and testing much simpler.
- GPU-enabled applications have both non-gpu and gpu-enabled versions installed. Typically, gpu-enabled versions are tagged with gpu in their module name, e.g., lammps/31Mar17_gpu is the GPU-enabled version of LAMMPS, while lammps/31Mar17 is the non-gpu version of LAMMPS.
- An exception to the above rule is that for licensed softwares like Abaqus, Ansys, and Matlab, a single module contains both non-gpu and gpu-enabled versions.
- A selection of GPU-enabled application containers from the Nvidia GPU Cloud (NGC) collection is installed.