Scientists from the Faculty of Chemistry are using the high performance computing (HPC) facility for performing cutting edge research in a variety of fields. Some examples include: understanding unique properties and behavior of materials and interfaces using the principles of quantum mechanics; deciphering the principles of cellular self-organization processes such as protein folding and protein-DNA recognition; understanding the physics of turbulence; modeling climatic and atmospheric phenomena; and much more.
The Chemistry HPC was established in June 2010. It consists of 602 dual socket compute nodes. All together, the farm provides 29,544 cores, about 153 TB RAM ranging from 4 to 16 GB per one core, and 6.0 PB of user disk space. There are 14 nodes with NVIDIA GPU accelerators, 12 of them are publicly available. The theoretical performance of the CPU nodes is about 2.1 PetaFLOPs, and 1.28 PetaFLOPs measured. The compute nodes are connected via InfiniBand network with fat tree topology, to allow for high-speed communication. The software installed includes advanced C and Fortran compilers, mathematical libraries, and a variety of commercial and homemade dedicated software suites that suit the user’ research needs.