What's new

India loses ground in supercomputing

Lankan Ranger

ELITE MEMBER
Joined
Aug 9, 2009
Messages
12,550
Reaction score
0
India loses ground in supercomputing

India's presence in a reputed global list of the world's fastest 500 supercomputers has been dwindling in recent years, with just two from the country finding a place in the latest list released in November.

The annual TOP500 list, updated in June and November, ranks the supercomputers based on their speed.

In the list that is dominated by the United States, China, Japan, the United Kingdom, Germany and France, India has mostly been relegated to the background in recent years.

The two supercomputers from India belong to Tata's Computational Research Laboratories and the Indian Institute of Tropical Meteorology.

In comparison, four Indian systems found a place in the November 2010 list: in addition to the two already mentioned, those of the Centre for Development of Advanced Computing (C-DAC) and the Indian government made it to the top 500.

Though the Indian Space Research Organisation had unveiled a supercomputer in May, said to be India's fastest in terms of theoretical peak performance, it has not been featured in the list, owing to a technicality.

It was in 2006 that the maximum number of supercomputers from India made to the list — 11 in June and 10 in November. Then, India ranked sixth in terms of supercomputing power; this November, it was placed fifth from the bottom in a list of 27 countries.

India's decline may mean that the kind of investments others, especially the Asian countries, are making in high performance computing (HPC) is not made in the country, says Horst Gietl, executive consultant, International Supercomputing Conference.

HPC's role

HPC plays an important role in exploring challenging problems requiring huge volumes of computation and data in such domains as climate modelling, bioinformatics, cosmology and molecular modelling, explains Sathish Vadhiyar, associate professor, Supercomputer Education and Research Centre, Indian Institute of Science, Bangalore.

The number of supercomputers in a country reflected the diversity, significance and magnitude of such problem-solving efforts. “India has definitely lost ground in the supercomputing domain, compared with China.”

This could be attributed to the lack of co-ordinated and integrated efforts between HPC systems researchers and application domain experts in exploring problems of large sizes on a very large number of processors. As a result, there were not enough domain experts interested in scaling their applications to use a large number of processors, and HPC scientists who could help to develop a comprehensive system software. However, the government had taken initiatives to address this situation, which could improve in the near future, he added.

In the world of supercomputing, power is associated with the number of calculations that could be performed a second by a machine. A teraflop equals one trillion floating point operations a second, and a petaflop, a 1000 trillion floating point operations a second.

The K Computer in Japan's RIKEN Advanced Institute for Computational Science, Kobe, with a speed of 10.5 petaflops, has been ranked first in the November list. In comparison, the Tata supercomputer, ranked 85th, has a maximum computational speed of 132.8 teraflops.

Sixteen supercomputers, including those that have found a place in the list, were featured in the June 2011 version of a bi-annual Indian list Dr. Vadhiyar has been preparing in recent years. (The ISRO supercomputer has been categorised separately in this list).

In the first version of this Indian list released in November 2008, 11 supercomputers were selected, with the entry barrier for being counted set at 900 gigaflops. In the list for June this year, the barrier was set at 3.11 teraflops.

So, does speed matter much more than numbers in the supercomputing domain? It is one indicator of the supercomputing resources at the disposal of a country. The more HPC research is done, the greater is the need for faster computers, says Dr. Gietl.

The Hindu : Sci-Tech / Internet : India loses ground in supercomputing
 
I appreciate ur comments !!!

Not expected from a chinese
 
India was never strong in supercomputing, so there is no such thing as the loss.
 
That is because negligence by the govt. Scientists repeatedly asked for necessary investments and power supply but govt is not interested. Companies like Wipro and some others offers high performance supercomputing. Anyway India has taken a project to develop high end supercomputer with in next few years.
 
That is because negligence by the govt.

What do you mean negligence by the Government?

What about private companies?

How did India go from having 11 in the top five hundred, to having just 2 today.... in just five years.
 
IMHO there hasn't been applications that justify continued investment in SC. I'd imagine the current horsepower is sufficient for our needs.
Anyway SC is pretty scalar, the architecture isn't that hard (well established) and all that is needed is adding more computing modules (Intel or Nvidia chips) to increase computing throughput.
 
IMHO there hasn't been applications that justify continued investment in SC. I'd imagine the current horsepower is sufficient for our needs.
Anyway SC is pretty scalar, the architecture isn't that hard (well established) and all that is needed is adding more computing modules (Intel or Nvidia chips) to increase computing throughput.

interconnects.

also, India wants to be making more nuclear weapons and/or maintaining the ones it has, but can't actually test them. the next best thing is a radiation hydrodynamics model on a computer. turns out that radiation hydrodynamics requires solving something like 9 coupled PDEs, which requires a supercomputer.

that's just to analyze the compression of the hydrogen and the resulting explosion; you need nuclear physics to analyze the energy that goes into the explosion, which is a whole different set of equations, that you have to solve for at the same time as the radiation hydrodynamics.
 
^your piece meal arguments are not adding to the discussion.
FYI GOI's focus is not on creating nuclear weapons, beyond a certain threshold nukes are a terrible waste of resources.
 
Back
Top Bottom