SYSTEMS
PARAM supercomputing power for Tanzania, too
The project is sponsored by ministry of external affairs, government of India through department of Information Technology, ministry of communications and Information Technology, government of India. Valued at Rs 3.74 crore, this system will be deployed at Dar-Es-Salaam Institute of Technology, Tanzania.
“The PARAM supercomputing facility will be used extensively for scientific and research applications programmes in addition to being a national data centre. This facility will support the government of Tanzania by way of capacity building in the areas of high performance computing and applications,” C-DAC Director General Rajan Joseph said.
The PARAM supercomputing cluster consists of indigenously developed PARAM Net III interconnect, software stacks, High Performance Computing Resource Management Engine (CHReME), High Performance Computing (HPC) portal, system software and tools along with storage. Additionally, system administration and application porting and parallelisation, training on HPC tools shall be provided.
“The system would cater to various applications in computational fluid, dynamics, weather forecasting, bioinformatics, finite element analysis, ocean modeling seismic and materials modeling. The project will be for duration of three years, wherein exchange programmes for scientists will be facilitated along with interactive workshops,” the C-DAC DG said.
The PARAM supercomputing clusters have been deployed in Russia, Singapore and Ghana. C-DAC is in discussion with other countries for establishing similar facilities. “We are holding talks with Armenia, Uzbekistan, Saudi Arabia, Myanmar and Nepal. We are in talks with Russia, which is using a sub-scale version of PARAM Padma to reinstall the new version of the supercomputer,” C-DAC, Pune director S P Dixit said.
C-DAC aims to break teraflops speed barrier by achieving petaflops computing speed by 2012 for its supercomputer Param Yuva, C-DAC director general Rajan T Joseph said.
“PARAM Yuva’s current speed is 54 teraflop. We intend to cross 1,000 Tflop. The aim is to scale up the operations ensuring same space is taken up by the unit without consuming additional power,” Joseph said. PARAM 8000 set the platform for a whole series of parallel computers, called the PARAM series, of HPC systems over the years, with PARAM 20000, or PARAM Padma, breaking the teraflop (thousand billion flops) barrier in 2002 with a peak speed of 1 Tflop.
The latest in the series is called PARAM Yuva, developed last year and ranked 68th in the TOP500 list released in November 2008 at the Supercomputing Conference in Austin, Texas, United States.