IBM to build supercomputer 20 times faster than the current No.1
04 Feb 2009
IBM has signed a deal with the Lawrence Livermore National Laboratory (LLNL) for a successor to its "Blue Gene" supercomputer, which will be code-named "Sequoia" and would have the computing power of 2 million laptops. This is also equivalent to the combined speed of the 500 fastest computers in the world today.
To put this into perspective, if each of the 6.7 billion people on Earth had a hand calculator and worked together on a massive calculation for 24 hours per day, 365 days a year, it would take 320 years to do what Sequoia will do in one hour, according to IBM.
IBM has set a 2012 target for the installation, including a goal of achieving 20 petaflops of computing power that the company hopes will far surpass the 2008 "Roadrunner" system, a 1.105-petaflop machine that the company installed at the Los Alamos Laboratory in 2008. (See: Roadrunner runs away with top supercomputer spot, breaks petaflop barrier).
Sequoia will include 1.6 million IBM Power processors housed in 96 racks the size of refrigerators, and occupy 3,422 square feet. It will be based on future Blue Gene technology to be delivered in 2011 and deployed in 2012. It also will have 1.6 petabytes of memory and 96,304 compute nodes. The system will contain nearly 100,000 computing nodes and draw six megawatts of power a year, enough to power 500 homes.
So far, IBM and the LLNL haven't disclosed how much the installation will cost, or to what specific purpose the new supercomputer will be put; the Roadrunner installation cost $100 million. IBM will also work on "Dawn," a 500 teraflop supercomputer. IBM said the new Sequoia system will be based on its POWER architecture.
Sequoia and Dawn will serve the National Nuclear Security Administration's (NNSA) tri-lab Advanced Simulation and Computing (ASC) program, which unites the scientific computing resources and expertise of Los Alamos, Sandia and Lawrence Livermore National Laboratories. The labs will run extremely complex simulations, including simulations that will ensure the safety and reliability of the nation's nuclear weapons stockpile, which could negate the need for real-world weapon testing.
Sequoia and Dawn will also be used for research into astronomy, energy, human genome science and climate change. Aside from the primary nuclear-related simulations, a 20 petaflop supercomputer would result in a 40-fold jump in the ability of scientists to monitor and forecast weather, according to IBM. This would allow forecasters to predict local weather events that affect areas 100 meters to one kilometer in size, down from their current 10-kilometer ability, IBM said. Weather forecasters could more easily and accurately predict tornadoes, which are local events generally confined to a few kilometres.
Additionally, according to researchers at Lawrence Livermore, a 20-petaflop machine would deliver 50-times improvement in scientists' capability to predict earthquakes and map out safe evacuation routes. This would let scientists predict an earthquake's effects on a building-by-building basis across an area as large as Los Angeles County.
