Nvidia: US steadily losing supercomputer race

Posted by Aharon Etengoff

A high-ranking Nvidia exec has warned that "lackluster" investment in supercomputing could cause the US to lose its technological edge over China.

"[Tianjin's] record-breaking Tianhe-1A supercomputer is no publicity stunt or state-sponsored play for mere pride. Delivering 1.4 times the performance of America's fastest supercomputer, it's a leading indicator of where computing is going," explained Andy Keane, Nvidia's GM of GPU Computing.


"[Remember], China is not chained to the anchor of legacy CPU-based computing. And while part of [Beijing's] innovation is in the networking interconnect to link processors together, their achievement is based on a new and more pragmatic vision about how to deploy hardware effectively.

"[Clearly], China has made the great leap into next-generation, hybrid supercomputing by using GPUs to drive far better efficiency and performance, more economically."

However, Keane emphasized that China's supercomputing achievements were hardly limited to a reconfiguration of traditional hardware paradigms.

"Educating the next generation of researchers and scientists is critical. But [unfortunately], computing capacity and infrastructure are being critically overlooked and under-invested in [America].

"Most supercomputers are already 2X over-subscribed at our current level of demand. And in this decade our level of science will see a 1,000-fold increase in its computational demands."



He also noted that there was hardly a major sector of the economy which wasn't "impacted daily" by supercomputing infrastructure.

"Design, imaging and simulations have become critical to transportation, consumer products, healthcare, energy, telecommunications and financial services - to name just a few.

"[Obviously], faster computation generates the innovations and competitive edge so critical to growth and improved standard of living."

As such, Keane urged policy makers and US-based investors to "consider the consequences" of falling behind in computing innovation over a longer horizon.

"A direct line can be drawn between the tools societies develop and the advances they make over time - in science and technology, in product development and innovation and in economic growth.

"[If] you trace that line - from the observations of ancient astronomers and mathematicians, and through the experimentation of the past two centuries - it's clear that the third and most vital wave of human innovation in this century is now being driven by computation.

"[So], ten years from now, if we find ourselves looking back at 2010 as the year American technology was enlisted to beat us at a game we invented - it will not be a failure of technology. [Rather], it will reflect a deficit of imagination."