AT is the essence of Google’s power? What does it take to predict tsunamis or what is the secret that helps decode the human gene? The answer to all is complex mathematical algorithms — which will now be studied and developed in Pune courtesy Tata Institute of Fundamental Research and Hewlett Packard who are setting up one of the largest supercomputers in India at the TIFR’s Computational Mathematics Laboratory (CML) here.
Just so we understand what is so ‘‘super’’ about the system: It’s a one teraflop supercomputer, ie, it can compute one trillion floating point operations in a second.
‘‘We can use the algorithms for many technological advances like nanotechnology. We can do high-end research in computational mathematics, engineering and medicine,’’ explained Dr Narendra Karmarkar, head of CML in Pune on Wednesday. The aim is to build Brand India in high-end R&D.
Karmarkar said this was the largest 64bit Open Systems high performance computing cluster in India. TIFR set up this $1 million facility in partnership with Hewlett Packard. TIFR has opted for the HP XC6000 clustering suite ‘‘whose simple interconnect design is useful for solving embarrassingly parallel problems’’, 156 Itanium 2 1.6GHz processors, which HP made jointly with Intel to increase the pace of performance at a lower cost and will operate it on Linux. This will be the largest roll-out of the Itanium Linux cluster.
On the choice of HP and Itanium, Karmarkar said, ‘‘We needed high performance computing deploying latest technology and HP fit the bill.’’
HP has a 32 per cent share in the global high performance computing market pegged at $3.8 billion, says Pallab Talukdar, director, enterprise marketing and alliance - customer solutions group of Hewlett-Packard India. CML will also contribute to the evolution of next generation HPC platforms and work closely with HP.
The lab’s ultimate aim? To have a petaflop supercluster, which can compute one quadrillion floating point operations in a second.