Just 3 Indian supercomputers in global Top 500 list
New ratings see both ‘ Made in India’ machines in list – at lower ranks AMD fuels world’s fastest computer – at 1.75 petaflops
The latest half yearly ratings of the world’s 500 fastest computers finds the number of India-based machines shrink from six to three since the last ratings in June. The two made-in-India platforms are still in the list, but the machines used by private commercial agencies is down to just one at an unnamed location.
Tata’s Eka supercomputer in Pune’s Computational Research Laboratories (CRL) remains the fastest Indian supercomputer, its top performance of 132.8 teraflops remaining unchanged. However shifting goalposts in high performance computing technology sees the Eka slip from no 18 to no 26 in the list.
The government-run Centre For Development of Advanced Computing (C-DAC) also headquartered in Pune, sees its Param Yuva machine at no. 137 in the latest ratings announced November 17. Its speed is in fact a bit faster than before -- 38.1 teraflops -- but its rank is now 137, down from 109 in June.
Both platforms are clusters, indigenously assembled, using Intel Xeon chip- fuelled nodes sourced from HP
The only other India based supercomputer is a 28.357 teraflop cluster also with HP hardware, and ranked no. 247.
There is an upset at the top of the list: A Cray XT5 “Jaguar” housed in US Department of Energy’s Oak Ridge facility has overtaken the IBM “Roadrunner” in the same department’s Los Alamos facility, clocking 1.75 petaflops or 1750 teraflops to IBM’s 1.04 petaflops. This also puts a machine fuelled by AMD at the top of the list -- indeed AMD’s Opteron processors fuel 5ive of the top ten machines. However Intel chips are to be found in400 of the top 500 supercomputers.
The full list can be found at www.top500.org
IndiaTechOnline Comment The small number of super computing platforms in India is a sobering reflection of scarce resources and inadequate deployment for mission critical as well as high performance computational challenges like climate modelling, natural resource exploration, disaster prediction and management. ( The US has 277 supercomputers, China has 21 in the latest Top 500 list). While the nation proudly touts top software talent, the list shows a skew between such skills and the requisite hardware support to stimulate and empower such talent . Clearly, there is a big gap between official word and deed, when it comes to arming Indian computer scientists with the tools to take on such challenging and often socially relevant tasks.