IBM today (Nov. 14) announced that the U.S. Department of Energy has awarded IBM contracts valued at $325 million to develop and deliver “the world’s most advanced ‘data-centric’ supercomputing systems” at Lawrence Livermore and Oak Ridge National Laboratories to advance innovation and discovery in science, engineering and national security.”

The world is generating more than 2.5 billion gigabytes of “big data” every day, according to IBM’s 2013 annual report, requiring entirely new approaches to supercomputing.

Repeatedly moving data back and forth from storage to processor is unsustainable with the onslaught of Big Data because of the significant amount of time and energy that massive and frequent data movement entails, IBM says, so the emphasis on faster microprocessors becomes progressively more untenable because the computing infrastructure is dominated by data movement and data management.

To address this issue, for the past five years IBM researchers have pioneered a new “data centric” approach — an architecture that embeds compute power everywhere data resides in the system, allowing for a convergence of analytics, modeling, visualization, and simulation, and driving new insights at “incredible” speeds.

IBM OpenPOWER Systems: greater than 100150 petaflops

IBM says the two Laboratories anticipate that the new IBM OpenPOWER-based supercomputers will be among the “fastest and most energy-efficient” systems, thanks to this data-centric approach. The systems at each laboratory are expected to offer five to 10 times better performance on commercial and high-performance computing applications compared to the current systems at the labs, and will be more than five times more energy efficient.

The “Sierra” supercomputer at Lawrence Livermore and “Summit” supercomputer at Oak Ridge will each have a peak performance of more than 100150 petaflops (compared to today’s fastest supercomputer, China’s Tianhe-2, with 33.86 petaflops) with more than five petabytes of dynamic and flash memory to help accelerate the performance of data-centric applications. The systems will also be capable of moving data to the processor, when necessary, at more than 17 petabytes per second (which is equivalent to moving over 100 billion photos on Facebook in a second) to speed time to insights.

The national laboratories offer researchers from academia, government, and industry access to time on their open computers to address grand challenges in science and engineering. Traditionally, the labs’ computers have been optimized to handle hardcore scientific problem solving, using techniques such as modeling and simulation. Increasingly, researchers are seeking help with projects in diverse domains such as healthcare, genomics, economics, financial systems, social behavior and visualization of large and complex data sets.

The Sierra and Summit systems will be used for the most mission-critical applications and represent the next major phase in the U.S. Department of Energy’s scientific computing roadmap to exascale computing, IBM says.

Exchanging data five to 12 times faster

Working with IBM, NVIDIA developed the advanced NVIDIA NVLink interconnect technology, which will enable CPUs and GPUs to exchange data five to 12 times faster than they can today. NVIDIA NVLink will be integrated into IBM POWER CPUs and next-generation NVIDIA GPUs based on the NVIDIA Volta architecture, allowing Sierra and Summit to achieve unprecedented performance levels. With Mellanox, IBM is implementing a state-of-the-art interconnect that incorporates built-in intelligence, to improve data handling.

IBM says it is already delivering data-centric technologies to clients, including IBM POWER8, IBM Elastic Storage Server, IBM Elastic Storage software, and IBM Platform Computing software, slated to be deployed and installed in the labs by 2017-2018.

In related news, the U.S. Department of Energy (DOE) announced approximately $100 million to further develop extreme-scale supercomputing technologies as part of a research and development program titled FastForward 2.

According to DOE, “FastForward 2 seeks to develop critical technologies needed to deliver next-generation capabilities that will enable affordable and energy-efficient advanced extreme scale computing research and development for the next decade. The joint project between DOE Office of Science and National Nuclear Security Administration (NNSA) will be led by computing industry leaders AMD, Cray, IBM, Intel and NVIDIA.”

UPDATE Nov. 14: Performance of Sierra and Summit supercomputers changed from “more than 100” to 150 petaflops, based on a clarification by an IBM spokesman, who explained that the IBM announcement was conservative.

IBM | The National Labs and the Rise of Data-Centric Computing