Now A Number-Crunching Giant From C-Dac

The incoming government will unveil Indias biggest achievement so far in hardware manufacturing a 100-gigaflops supercomputer developed by the Pune-based Centre for Development in Advanced Computing (C-DAC).

Billed as one of the most powerful number-crunching machines in the world, the Param-100 model is said to be on par with the high-end machines of global hardware giants like Cray, IBM and Fujitsu.

The Param-100, which is slated for launch in March, is being given final touches by C-DAC engineers. The machine has been developed with the help of a Rs 48 crore government grant, official sources said.

The machine will be offered to the Indian Meteorological Bureau (IMB) as a replacement for its 10-year-old Cray supercomputer, which was purchased from the United States under stiff end-user conditions. Silicon Graphics, which now owns Cray, is in the race to replace the Indian Meteorological Bureaus machine.

 

Sources said the latest Param machine is based on open frame parallel architecture and will be optimised to run most high-end scientific applications. Apart from weather forecasting, it can also be used for industrial design, satellite data interpretation, complex imaging and defence purposes.

The Param-100 is expected to cost over Rs 50 crore. The machine marks C-DACs third hardware milestone.

The first was the development of a 2-gigaflop machine and the second, a 10-gigaflop computer. Sources said the new machine could be upgraded to 1,000 gigaflops, if needed.

C-DAC has some 30 installations of its 2-gigaflops machines so far, mostly in Indian public sector scientific institutions.

The organisation has not yet achieved much marketing success, even though its machines are price-competitive.

One reason for this is that the parallel architecture in the Param computers requires special software and constant debugging.

They are not as user-friendly as IBM or Cray machines, but over the years, some improvements have been made, said a source in a defence laboratory.