Big problems – figuring out how the brain works, gauging the
effectiveness of H1N1 flu drugs, recreating the universe’s origins – demand big
computer smarts. We’re talking teraflops – trillions of floating point
instructions per second.
The November issue of Wired
magazine features its choices for the world’s
biggest, baddest supercomputers/super-clusters. While “low performance
computing” and the cloud are the topic du jour in
grid and high-performance computing circles, let’s face it: For raw excitement and computing excitement,
these more traditional “super simulators” still sizzle. By order of size:
Ranger
Texas Advanced Computing Center, Austin, Texas
Processors 15,744
Peak teraflops 504
Replicates the interactions of the H1N1 flu virus with the human body and
the effect of various over-the-counter drugs.
Earth Simulator
Japan Agency for Marine Earth Science and Technology, Yokohama, Japan
Processors 1,280
Peak teraflops 131
Runs global climate simulations. A recent upgrade tripled its speed, meaning
more accurate and timely predictions.
Discover
NASA Goddard Space Flight Center,
Greenbelt, Maryland
Processors 11,032
Peak teraflops 111
Reproduces Earth’s paleolithic-era
climate and uses it to test new models for predicting weather patterns.
Blue
Brain
Ecole Polytechnique Fédérale de Lausanne, Switzerland1
Processors 8,192
Peak teraflops 22.4
Simulates the interactions among neurons in the neocortex — the brain’s
reasoning center — to learn how thinking works
Visualization Cluster
Hayden Planetarium, New York City
Processors 300
Peak teraflops 5.54
Re-creates the universe. When the visitors leave, scientists from around the
world take over the Planetarium to play what-if.
True cyber speed freaks will want to keep score of the
world’s baddest big iron at http://www.top500.org/
