LANL has always been an early adopter of transformational high performance computing (HPC) technology. For example, in the 1970s when HPC was scalar; LANL acquired the first Cray-1 vector supercomputer. When HPC was vector; LANL acquired the first TMC CM-5 massively parallel supercomputer; the first #1 on the TOP500 list. In the 2000s, HPC was distributed memory; LANL and IBM built Roadrunner, the first hybrid supercomputer and the first supercomputer to attain a sustained petaflop/second.
All articles in the IBM Journal of Research and Development, Volume 53, Number 5, 2009, are on hybrid computing systems, focusing on Roadrunner. Paper 8, The reverse-acceleration model for programing petascale hybrid systems, is authored by LANL CCS-1 staff Scott Pakin, Mike Lang, and Darren Kerbyson.
Open Science Press Releases in LANLtoday News
In a test run on May 27, 2008, the Roadrunner supercomputer, built by IBM with funding from the National Nuclear Security Administration (NNSA) for Los Alamos National Laboratory, achieved a long-sought supercomputing goal: performing more than a thousand trillion operations per second, or petaflop/s.
A “flops” is an acronym meaning floating-point operations per second. One petaflop/s is 1,000 trillion operations per second. To put this into perspective, if each of the 6 billion people on earth had a hand calculator and worked together on a calculation 24 hours per day, 365 days a year, it would take 46 years to do what Roadrunner would do in one day.
Roadrunner is the first supercomputer to use a hybrid processor architecture, which is based on both Opteron X64 processors from Advanced Micro Devices (AMD) and the IBM Cell Broadband Engine™ (Cell BE) processing elements.
Roadrunner will be housed at NNSA’s Los Alamos National Laboratory. The laboratory worked collaboratively with IBM, the manufacturer, for six years to deliver a novel computer architecture that can meet the nation’s evolving national security needs. The result has redefined the frontier of supercomputing, not only by crossing the one petaflop threshold, but also by introducing a new paradigm for the future.