Feeds

IBM wins US Gov super computer deal

Bags $290m for two monsters

  • alert
  • submit to reddit

Next gen security for virtualised datacentres

The US Department of Energy (DOE) has awarded IBM a $290 million contract to build the two fastest supercomputers in the world with a combined peak speed of 460 trillion calculations per second (teraflops).

The first supercomputer (ASCI Purple) will be used to simulate the ageing and operation of US nuclear weapons. The second system (Blue Gene/L) will be used in scientific research, in areas such as predicting global climate change, earthquake prediction and astrophysics.

ASCI Purple (successor to IBM's ASCI White, and more than eight times more powerful) will be the world's first supercomputer capable of 100 teraflops - or almost three times faster than the most powerful computer in existence today, according to IBM. ASCI Purple will consist of a cluster of POWER-based IBM eServers and IBM storage systems. ASCI Purple will be delivered in stages with the first IBM eServer systems arriving next year.

The ASCI Purple system is to be powered by 12,544 POWER5 microprocessors, IBM's next generation microprocessor. These processors will be contained in 196 individual computers with a total memory bandwidth of 156,000 GB. All of the computers are interconnected via a super-fast data highway with a total interconnect bandwidth of 12,500 GB. ASCI Purple will run IBM's AIXL operating system. The combined system will contain 50 terabytes of memory and two petabytes of disk storage.

IBM will integrate special autonomic self-managing features into the new supercomputer, allowing for ease of administration and greater system reliability. The new POWER5 processors will detect and recover from errors without the intervention of technician, IBM promises. When the system detects repeated errors it will move the workload to another part of the machine.

IBM hopes to bring some of these autonomic features developed for ASCI Purple into its future commercial systems. ASCI Purple will serve as the primary supercomputer in the US Department of Energy's Advanced Simulation and Computing Initiative, commonly known as ASCI.

Blue Gene high

The second supercomputer, a research machine called Blue Gene/L, will employ IBM semiconductor and system technologies based on new architectures being developed during the ASCI Program.

When completed, Blue Gene/L will have the peak performance of 360 teraflops with 130,000 processors running Linux.

IBM said the supercomputer will have the capability to process data at a rate of one terabit per second, equivalent to the data transmitted by ten thousand weather satellites. The supercomputer will be used by the three National Nuclear Security Administration laboratories (Los Alamos, Sandia and Lawrence Livermore) as well as other DOE laboratories and members of the ASCI University Alliance.

Blue Gene/L will develop and run a broad suite of scientific applications including the simulation of very complex physical phenomena, such as turbulence, the effect of radiation on DNA strands, and behaviour of high explosives. Scientists are also looking forward to applying the monster machine to problems in astrophysics, such as how the orbits of binary stars with a high mass ratio become unstable and merge. 3D seismic and acoustic wave propagation calculations on Blue Gene/L could be used for numerous diverse projects, including earthquake hazard analysis, oil exploration.

"Blue Gene/L represents a leap forward in the ASCI strategies of accelerating computing development," said Mark Seager, assistant director for Advanced Technologies for Livermore's Computation Directorate. "It's like having an electron microscope when all the other scientists have a magnifying glass." ®

Related Stories

IBM racks up 128 CPUs with P655
16,000 Hammers in Sandia supercomputer
UK boffins get supercomputer boost
IDC takes a second pass at supercomputer rankings
NEC captures supercomputing crown
IBM builds absolutely super computer
World's fastest supercomputer goes down a bomb

External Links

TOP500 List Supercomputing Sites

Build a business case: developing custom apps

Whitepapers

Best practices for enterprise data
Discussing how technology providers have innovated in order to solve new challenges, creating a new framework for enterprise data.
Implementing global e-invoicing with guaranteed legal certainty
Explaining the role local tax compliance plays in successful supply chain management and e-business and how leading global brands are addressing this.
Advanced data protection for your virtualized environments
Find a natural fit for optimizing protection for the often resource-constrained data protection process found in virtual environments.
How modern custom applications can spur business growth
Learn how to create, deploy and manage custom applications without consuming or expanding the need for scarce, expensive IT resources.
High Performance for All
While HPC is not new, it has traditionally been seen as a specialist area – is it now geared up to meet more mainstream requirements?