Colloquy: High-performance computing

November 6, 2013

Colloquy: Editor’s Column

BP will invest US$100 million in high-performance computing over five years, and part of that is being spent on a new facility at the Westlake campus in Houston, housing the world’s largest super computer for commercial research – the Center for High-Performance Computing. The CHPC will serve as a worldwide hub to process and manage geophysical and geological data across BP’s worldwide portfolio. BP said its computing needs are 20,000 times greater today than they were in 1999.

Jackie Mutschler, BP’s head of upstream technology, said “BP’s investment in this new supercomputing center not only highlights the increasingly high-tech nature of today’s global oil and gas industry, it underscores our company’s long-held belief in the vital role technology plays— and will continue to play—in solving the world’s biggest energy challenges.”

The three-story facility provides 110,000sq ft of space. The new supercomputer will have memory of 1000 terabytes (1 petabyte) and disk storage space for 23.5 petabytes – equivalent to more than 40,000 average laptop computers or 1.5 million 16GB iPads, which, laid sideto- side, would stretch from Washington D.C. to Philadelphia.

By the end of the year, BP’s computing center will hold 6000 computers. BP worked with HP and Intel to grow its computing power to over 2.2 petaflops, nearly doubling the company’s capability in one year.

In computing, FLOPS (for Floatingpoint Operations Per Second) is a measure of computer processing speed and performance. A kiloflop=103; megaflop= 106; gigaflop=109; teraflop=1012; and a petaflop is a quadrillion (1015) floating point operations per second.

IBM-built Roadrunner was the first supercomputer to reach 1 petaflop. It cost $120 million to develop at Los Alamos National Laboratory, and had a lifetime of just five years. It was the world’s fastest supercomputer until June 2009 but was shut down in early 2013 because of inefficient energy consumption. Roadrunner required 2345 kW to hit 1.042 petaflops (444 megaflops/watt).

BP’s new facility features improved electrical and cooling systems that reduce power consumption by 30% over the current computing facility.

While 2.2 petaflops is a massive facility by almost any measure, news broke this past June about a new Intel-based system in China, Tianhe-2 (“Milkyway-2”), that reached 54.9 peta-flops. This is more than twice the speed of any system in the US, said Jack Dongarra, a professor of computer science at the University of Tennessee. The Chinese government spent $290million building the new system, which replaces Tianhe-1, dubbed the world’s fastest in 2010. OE

Image Caption: BP’s new Center for High-Performance Computing opened in October.
Photo: BP.