World's top supercomputer this time is...

FRANKFURT, GERMANY: It's Sunway TaihuLight from China again. Yes, it has been ranked officially as the fastest supercomputer in the world.

The chops: a theoretical peak performance of 125 petaflops, as many as 10,649,600 cores, a Linpack mark of 93 petaflops and a massive 1.31 petabytes of primary memory.

It has beaten US contenders as well as the previous home-champ
Tianhe-2 (by a factor of three).

But what makes things more interesting is that China's top wrestlers in the supercomputing ring are elbowing out the ones from the US with speeds five times faster. And when that happens with indigenous chips (Intel is no more under the hood after a political turn of events causing an embargo), well, is that salt to the wound?  ShenWei processors instead of XeonPhi, yeah.

There's more happening in the supercomputer picnic happening in the meadows of the 2016 International Supercomputing Conference here in Frankfurt. More, that puts the arclights on applications, actual world use and commercial developments, besides all the hardware face-off.

Notable super-computer name Cray Inc. announced about new contracts for some of its supercomputers. There's one from the Academic Center for Computing and Media Studies (ACCMS) at Kyoto University in Kyoto, Japan for a Cray XC40 supercomputer, featuring the Cray DataWarp applications I/O accelerator, and a Cray CS400 cluster supercomputer.

The Centre for Modelling & Simulation (CFMS) in Bristol, United Kingdom  also purchased a Cray CS400 cluster supercomputer, and the U.S. Department of Energy’s Argonne Leadership Computing Facility at Argonne National Laboratory has given a contract for the recently launched Cray Urika-GX system.

As further explained, the researchers and scientists at Kyoto University’s highly regarded ACCMS will use its new Cray systems to conduct research and development related to the advanced use of information technology infrastructure and information media. ACCMS will use its Cray XC40 supercomputer and Cray CS400 cluster supercomputer for applications that require scalability and high sustained performance, as well as applications that require large capacity for multiple jobs.

Professor Hiroshi Nakashima, Chair of ACCMS’s Supercomputing Service Committee added, “Our new Cray systems will be available to universities and research institutions across Japan, and will play an important role in several nationwide, joint research projects. Cray’s advanced supercomputing technologies will play an integral role in support of our mission.”

The Centre for Modelling & Simulation is currently installing the first-ever Cray supercomputer in the city of Bristol and it is being informed that the Cray CS400 system will serve as the primary high performance computing system at CFMS.

"With modelling and simulation at the core of product development, providing a technology platform that enables rapid innovation and supports the acceleration of engineering design capability is central to our vision,” said Sam Paice, Chief Operating Officer of CFMS.

Fujitsu on the other side, has waved buh-bye to SPARC and announced to use 64-bit ARMv8 cores in the upcoming Post-K exascale supercomputer and it is being speculated that when operational, the Post-K will not only make the world's fastest known computer but also the first ARM-powered supercomputer.

Coming to the TOP500 list again, the SGI ICE X system, a 5th generation ICE system, has been ranked as the highest performing commercial supercomputer with 6.7 petaflops (peak).

Machine learning is also stealing some attention in HPC world this time with Boston Limited bringing out Boston ANNA Pascal, a new NVIDIA Tesla GPU-based solution at ISC 2016. It is explained that the NVIDIA Pascal architecture is purpose-built to act as the engine of computers that learn, see, and simulate our world and hence, all set for the title of the ‘world’s fastest deep learning appliance’.

It can help the research community and the industry a lot going by what is claimed;  and help solve many big data problems such as computer vision, speech recognition, and natural language processing.

Things are opening up a lot in many ways if we pause to think.

SUSE and Intel have come together at ISC 2016 and the company will distribute SUSE Linux Enterprise Server for High Performance Computing as an option for the recently introduced Intel HPC Orchestrator, an Intel-supported HPC system software stack.

Notably enough, the SUSE Linux Enterprise Server is the first commercially supported Linux to be included as an operating system option with Intel HPC Orchestrator system software stack, and the companies are expected to show joint support for the combined solution.

Dell too announced  some advancements to its high performance computing (HPC) portfolio, like the availability of new Dell HPC Systems, and vertical-specific solutions with new systems for specific science, manufacturing and analytics workloads.

ISC 2016 is indeed happening.