Amrita School of Biotechnology sets up supercomputer built on NVIDIA GPUs
Amrita School of Biotechnology has setup a High-Performance Computer (HPC) system based on the latest models of Nvidia Corp's Graphics Processing Unit (GPU) and started its operation. The facility allows intensive computations on Graphics Processing Units (GPUs). GPUs are effectively parallel multi-core systems. Such architectures allow execution of many concurrent threads of program at the same time to get better performance throughput. The advantage is also to maximize performance at low cost and lower power requirements.
The supercomputing facility, called "manas" has an initial capacity of 5.579 Teraflops on a single multi-CPU system. This was partly supported by NVIDIA Pvt. Ltd. based at Pune, India. The system uses NVIDIA's Kepler and Tesla GPUs to improve the computational capacity.
The working goal of this new facility is to drive simulations and data crunching during large-scale simulations for understanding the brain and for computer-aided drug-discovery. The Computational Neuroscience lab has been developing and running large-scale spiking and biorealistic models of cerebellum.
One of the key goals is also to encourage tighter collaboration among computing system architects, programmers and researchers, providing an exchange between sciences and informatics. Several tools used at various labs are open-source and hence such facilities add to advance hypothesis-making science via computational modeling. As part of the curriculum, students of M.Sc. Bioinformatics and Computational Biology receive some training on CUDA, an architecture developed by NVIDIA for making useful algorithms on such platforms.
The facility will also support the role of the School as an NVIDIA CUDA Teaching Center at Amrita University. The CUDA-based supercomputing unit will enhance the existing CPU-based supercomputing clusters, ATMA and SOMA already available at the School.
Computational Neuroscience lab at the School of Biotechnology is also taking part in High Performance Computing projects hosted by Super Computing Applications and Innovation, CINECA, Italy jointly with University of Pavia.
February 19, 2014
School of Biotechnology, Amritapuri