by Anton Shilov
11/18/2011 | 09:13 AM
IBM this week announced its next generation supercomputing project, Blue Gene/Q that will provide an ultra-scale technical computing platform to solve challenging problems facing engineers and scientists at faster, more energy efficient, and more reliable rates than ever before. Blue Gene/Q is expected to predict the path of hurricanes, analyze the ocean floor to discover oil, simulate nuclear weapons performance and decode gene sequences.
The third generation in the Blue Gene family of supercomputers, Blue Gene/Q operates at an order of magnitude faster than previous systems, deploying 16 multi-processing core technology and a scalable peak performance up to 100PFLOPS. The new IBM PowerPC A2 processing architecture plays a key role in delivering performance. Each processor includes 16 compute cores (up from four used with Blue Gene/P, the previous system) plus a core allocated to operating system administrative functions and a redundant spare core.
Designed with a small footprint and low power requirements, Blue Gene/Q was ranked as the number-one most energy-efficient supercomputer in the world by the Green500 (June 2011). It provides low latency, high performance runs that simplify tracing errors and tuning performance, all based on an open source and standards-based operating environment, according to IBM. Engineered with fewer moving parts and built in redundancy, Blue Gene/Q has proven to be a class-leader in reliability. Blue Gene/Q’s combination of high reliability and energy efficiency makes it an economic supercomputing solution, with fewer failures translating to faster times for more sound solutions.
Blue Gene/Q incorporates architectural advances that contribute to the system’s outstanding performance and help simplify programming. For example, hardware-based speculative execution capabilities facilitate efficient multi-threading for long code sections, even those with potential data dependencies. If conflicts are detected, the hardware can backtrack and redo the work without affecting application performance. Hardware-based transactional memory helps programmers avoid the potentially complex integration of locks and helps eliminate bottlenecks caused by deadlocking - when threads become stuck during the locking process. Hardware-based transactional memory helps to deliver efficient and effective multi-threading while reducing the need for complicated programming.
“Completing computationally intensive projects for a wide variety of scientific applications that were previously unsolvable is not just possible - it is now probable. 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,” said Brian Connors, vice president of technical computing at IBM.
When Blue Gene/Q is fully deployed in 2012 at Lawrence Livermore National Laboratory, the system, named “Sequoia”, is expected to achieve 20PFLOPS at peak performance, which is approximately two times higher compared to the currently deployed K computer. The capabilities this system represents will help ensure United States leadership in high performance computing (HPC) and the science it makes possible. Moreover, Blue Gene/Q is expected to become the world’s most power-efficient computer, churning out 2GFLOPS per watt.
LLNL, a premier multidisciplinary national security laboratory for DOE’s National Nuclear Security Administration (NNSA), applies some of the world’s most powerful supercomputers to maintaining the nation’s aging nuclear deterrent without testing, as well as such challenges as grid and network management, energy research and climate change. IBM will deploy 96 racks beginning as early as December of this year.
“It is this emphasis on reliability, scalability and low power consumption that draws the interest of NNSA to this machine and its architecture. This machine will provide an ideal platform to research and develop strategies to assure that our most challenging codes run efficiently on multi-core architectures. Such capabilities will provide tremendous information in formulating our code development strategy as we face the challenges of exascale simulation and advance the state of the art in simulation science, advances necessary to ensure our nation's security without nuclear testing,” said Bob Meisner, head of NNSA's advanced simulation and computing program.
Announced earlier in 2011, Argonne National Laboratory (ANL) will also implement Blue Gene/Q to stoke economic growth and improve U.S. competitiveness for such challenges as designing electric car batteries, understanding climate change and exploring the evolution of the universe. The 10PFLOPS system, named “Mira”, will provide a strong science and technology engine that will fuel national innovation. Argonne is one of the DOE’s oldest and largest labs for science and engineering research, located outside of Chicago.
A commitment to energy efficiency is essential as IBM continues to point forward towards exascale computing. An ExaFLOPS, a million trillion calculations per second, is 1000 times or three orders of magnitude faster than today's petaflop-class systems. Exascale computing has the potential to address a class of highly complex workloads that have been beyond traditional reach, not just due to their sheer size, but because of their inherent uncertainties and unpredictability – areas such as hurricane prediction and climate modeling. Blue Gene/Q represents the next step in this HPC evolution.