IBM to Deploy 20 Pflops Blue Gene/Q Supercomputer at LLNL

IBM may soon take back the title of world's fastest supercomputer with an upcoming Blue Gene/Q machine that, when installed, should be able to deliver almost twice the performance of today's fastest HPC system.

The supercomputer, names “Sequoia,” is expected to be fully deployed in 2012 at Lawrence Livermore National Laboratory (LLNL) and should achieve 20 petaflops peak performance.

In addition to its impressive number crunching capabilities, IBM also expects its system to become one of the most power-efficient supercomputers in the world, churning out no less than 2 gigaflops per watt.

These advancements in energy efficiency and computing power were made possible by IBM's new PowerPC A2 processing architecture which adds 16 computing cores to each processor installed, compared to the four cores used in its previous machine, the Blue Gene/P.

Together with these 16 cores dedicated to providing the computing power of the CPU, IBM's chip also sports a special core allocated to operating system administrative functions and a redundant spare core.

Sequoia's performance is also improved thanks to a series of hardware-based speculative execution capabilities, meant to facilitate writing multi-threaded code for the machine, as well as to the introduction of hardware-based transactional memory.

“Completing computationally intensive projects for a wide variety of scientific applications that were previously unsolvable is not just possible - it is now probable,” said Brian Connors, VP of technical computing at IBM.

“IBM’s historic role in developing the supercomputers that provide the power behind critical applications across every industry has uniquely positioned us to provide reliable supercomputing at the highest level,” concluded the company's rep.

Right now, the fastest supercomputer in the world is the Japanese K-Computer installed at the Riken Advanced Institute for Computational Science in Kobe, Japan which is capable of delivering 10.51 petaflops.

The system is comprised of 864 racks including a total of 88,128 interconnected Fujitsu Sparc64 VIIIfx CPUs featuring eight cores and running at 2GHz.