IMSc is relatively a small institute with an international reputation in the fields of mathematics, theoretical physics and theoretical computer science and we are proud to mention that ours is also an important centre for high-performance computation in India. IMSc is in the 27th rank of the systems in terms of Rmax (Linpack Benchmark performance) in the Indian supercomputers list. Our latest super computers are Annapoorna (SGI Altix ICE8200 EX Cluster) developed in the year 2010 and Dell R720 Rack Server developed in the year 2014.
IMSc's supercomputers history are given below in the chronological order
CSC - Cluster Super Computer (1998)
This is the first cluster built at IMSc in the year 1998. This system had 16 nodes, Pentium – III Processors. The system was used for the simulations, numerical calculations and home grown codes.
NCSC - New Cluster Super Computer (2003)
In the year 2003, The New 9 node Cluster Super Computer (NCSC) was purchased with Intel Xeon 2.4GHz, home built for the use of simulations and lattice gauge theory. Our institution has a tradition of naming computer systems by peaks/ranges.
KABRU (2004)
Kabru is a Home-built, 144 nodes Xeon DP Linux cluster. This system uses a 2.4GHz Intel xeon processors, Dolphin SCI 3D torus Interconnect and linux to provide a sustained speed of 959 gigaflops. This was used to handle theoretical physics research which dealt primarily with large scale simulations in teh filed of the lattice gauge theory. With a sustained performance of 1 TF (on October 2004), the second fastest supercomputer in India and the fastest supercomputer in India belonging to an academic institution at that time. Till 2012 Kabru was a part of the Garuda grid that links 45 institutions across 17 cities for nationwide computational nodes and mass storage.
In 2004,Kabru was listed as #257 in the world's top 500 super computers.
Vindhya (2006)
The Cray XD1 was purchased in the year 2006 for the use of theoretical physics research computations. This was an entry-level supercomputer range made by M/s. Cray Inc with AMD Opteron 64 bit CPUs and utilizes the Direct Connect Architecture over Hyper Transport to remove the bottleneck at the PCI and contention at the memory. The MPI latencyis 1/4 that of infiniband and 1/30 that of gigabit thernet.
Aravalli (2008)
Aravalli is a 4 node Sun Micorsystems AMD cluster based on sunfire X4600 M2 nodes. Each node having 32 Core AMD Optron 2.4 GHz Processor. This is purchased under the project.
Annapurna (2010)
Annapurna is a factory-integrated computational SGI Altix ICE8200 EX Cluster with 1024 cores of Intel Nehalem 2.93 GHz chips, 1.5 tera byte (TB) memory and storage of 30 TB with a peak speed of 12 teraflops (TF) Annapurna.
It is used for Simulations and numerical calculations, predominantly in the areas of statistical mechanics and condensed matter physics, lattice gauge theory, biophysical simulations, bioinformatics, chaos, complex and dynamical systems. Symbolic manipulation languages and various statistical packages used extensively in General Relativity, Gravitation and High energy physics. In theoretical computer science, computational power for the following types of problems: simulation of algorithms for gauging empirical performance, as for instance, in the Algorithms WorkBench), software for computational geometry and in cryptography.
At peak speeds of 12 teraflops Annapurna was ranked as the third fastest cluster among the broad-based scientific institutions in India after those at the IISc Bangalore and Tata Institute of Fundamental Research, Mumbai.
In 2010 it is the seventh fastest high performance computing machine in India.
Satpura (2012)
Satpura is a quad-way symmetric multi processing super micro rack server with AMD opteron 2.1 GHz, 12 core processor, 192 GB DDR3 memory developed by IMSc in 2010. It is used by the parallel users ..................
Nandadevi (2013)
Our latest HPC Hybrid System is Dell R720 Rack Server with 49 nodes classified into 10 TF cluster , Serial computing , GPU computing , Intel MIC computing , Luster server and Dual controller based 25 TB parallel storage. It is developed for Parallel computing, serial computing, montecorelo simulations, GPU computing etc.