The re-entry of Cray generates much heat in the Indian supercomputer industry
MAX MARTIN and N RAGHURAM
cray, the American supercomputer which was denied to India seven years ago is now hitting the market. The new Crays, offered by Silicon Graphics-Cray Research tie-up, are much more advanced than the Cray- xmp , the limited power supercomputer the us grudgingly sold under strict conditions and placed under direct us security till 1994 at the National Medium Range Weather Forecasting Centre ( nmrwfc ), New Delhi. "We find a potential market in India. And the us government has relaxed it embargoes on selling supercomputers," says Pradeep Marwaha, business development manager of Silicon Graphics-Cray Research.
As the us had blocked Cray's use in space and nuclear research, Bhabha Atomic Research Centre ( barc ) and Advanced Numerical Research and Analysis Group, a defence research body developed their own supercomputers for strategic use. Indian Institute of Science and National Aerospace Laboratories ( nal ) followed suit. "Now the us does not have just market interest; they don't want the supercomputer industry to develop here," says V P Bhatkar, executive director of the Centre for Development of Advanced Computing ( c-dac ), a scientific society under the Department of Science and Technology ( dst ).
Cray's re-entry means serious heartburn to Indian supercomputer developers, a dozen-odd institutions who assemble their own machines. Since 1991, c-dac has sold 30 of its own programme, the param machines, four of them abroad. Still, Bhatkar questions the wisdom of opening up a 'high-risk market' -- very limited market wher e technological innovations get outdated fast. Bhatkar says, "Europe uses their own supercomputers, the Japanese were not allowed to compete in the us, so why do we allow them?"
Ironically, even barc is going for a Cray. H K Kaura, head of barc 's computer division confirms that they have been planning to buy a Cray for a long time, "But it is not true that we have already placed an order," he says. Expounding upon why barc needs a Cray despite having their own supercomputer Kaura says, "We will need Cray for high-speed jobs. Our own parallel processing computers are there, but they will have to be modified for these applications".
Scientists sound a bit amazed. Says C K Raju, a former c-dac scientist, presently working at the National Institute of Science Technology and Development Studies, New Delhi, "If nal and barc are really buying Cray machines, I would say it is extremely strange, especially since they have had their own development programme for the last 10 years. I fear that the day they get Cray they will neglect their efforts at indigenous development. And once this happens, it is very likely that the latest Cray supercomputers would be refused to us once again. This danger is always there."
Perhaps, only the strategic areas will now remain aloof from foreign supercomputer entry. That means a lot of heat in the market. c-dac , after losing a few key supercomputer engineers to Silicon Graphics, now plans to become a company with greater flexibility and market muscle. c-dac 's latest param model is of two gigaflops (one giga = 1000 million), theoretically 10 times powerful than the j -90 -- one of Crays most prestigious models. But param is not a machine that you can buy off the shelf and start using. The end user will have to tailor his own programme.
"The difference is like a product and a prototype," quips Lokesh Bhasin, a former c-dac engineer now with Silicon Graphics-Cray Research. Bhatkar retorts: "It is the colonial, slavish Indian mentality to say anything western is better."
The market seems to wait for none. nmrwfc has plans to buy an advanced Cray, according to reliable sources. This comes after arranging certain weather forecasting programmes of the center's Cray- xmp to three Indian supercomputers -- param , Flosolver of nal and Anupam of barc . Says Bhasin, who was part of the arranging programme for param two years ago, "It did not give exact results, the speed up was not proper, stable performance could not be ensured." Meanwhile, the Cray there was giving a 97.5 per cent results. Bhatkar finds the arguments out of context. "It was a programme meant for Cray; still we could meet the bench marks." He shows a test result showing param running the programme gradually, drastically cutting down the time required for the programme.
We are a voice to you; you have been a support to us. Together we build journalism that is independent, credible and fearless. You can further help us by making a donation. This will mean a lot for our ability to bring you news, perspectives and analysis from the ground so that we can make change together.
.png)
Comments are moderated and will be published only after the site moderator’s approval. Please use a genuine email ID and provide your name. Selected comments may also be used in the ‘Letters’ section of the Down To Earth print edition.