Mira (supercomputer)
| Company type | Supercomputer |
|---|---|
| Industry | High-performance computing |
| Founded | 2012 |
| Headquarters | Argonne, Illinois , United States |
Key people | IBM (developer), Argonne National Laboratory (operator) |
| Products | Blue Gene/Q supercomputer |
| Owner | United States Department of Energy, National Science Foundation (partial funding) |
Mira is a decommissioned petascale Blue Gene/Q supercomputer that held the 11th position on the TOP500 ranking as of November 2017, having originally appeared in June 2012 at 3rd place. It achieved a performance of 8.59 petaflops on the LINPACK benchmark while operating with an energy consumption of 3.9 MW. Developed by IBM for the Argonne Leadership Computing Facility at Argonne National Laboratory, Mira received backing from the United States Department of Energy and partial financial support from the National Science Foundation. The system facilitated scientific investigations across fields such as material science, climatology, seismology, and computational chemistry. At its inception, it was allocated to sixteen projects chosen by the Department of Energy.
The creation of the Argonne Leadership Computing Facility was authorized under the America COMPETES Act, enacted during the administrations of Presidents Bush and Obama, as part of U.S. initiatives to maintain competitiveness with technological advancements like China's Tianhe-1A, which dominated global supercomputing rankings from October 2010 through June 2011. Mira was one of three American petascale supercomputers deployed in 2012, alongside IBM Sequoia and Blue Waters.
IBM has not publicly revealed the exact cost of Mira's construction; preliminary figures estimated approximately US$50 million, whereas Argonne National Laboratory cited acquisition through a grant totaling US$180 million. IBM emphasized Mira's computational power by illustrating that if every individual in the United States performed one calculation per second, it would require almost a full year to equal just one second of Mira’s processing capacity.
One of the applications
Researchers at Argonne utilized Mira to create a hybrid material demonstrating superlubricity on a macroscale for the first time. Their simulations involved up to 1.2 million atoms under dry conditions and up to 10 million atoms in humid settings, employing the LAMMPS software for reactive molecular dynamics modeling. To address performance limitations within LAMMPS' ReaxFF module, they introduced OpenMP threading, substituted point-to-point MPI communications with MPI collective operations, and implemented MPI I/O techniques, resulting in a twofold increase in execution speed.
The team is currently seeking patent protection for this hybrid material, which shows promise for applications in dry environments such as computer hard drives, wind turbine gear systems, and rotating seals used in micro- and nanoelectromechanical devices.