Thunderbird Linux supercomputer hits 53 teraflops

The Thunderbird Linux cluster contains nearly 9,000 CPUs

A monster Linux cluster loaded with almost 9,000 CPUs has been rated sixth in the Top 500 Supercomputer global rankings.

The Thunderbird Linux cluster was developed by Sandia National Laboratories in collaboration with Dell and Cisco.

Maintaining its sixth position in the Top 500 Supercomputers meant achieving an improved overall performance of 53 teraflops, an 18.5 per cent increase in efficiency from last year's performance.

The system comprises 4,480 commodity servers linked with an InfiniBand message-passing interconnect, making Thunderbird the largest cluster of its type in the world.

The improvement compared with last year was achieved by switching to OpenFabric Enterprise Distribution, a Linux-based open source software stack qualified by the OpenFabrics Alliance to operate with multi-vendor Infiniband hardware.

The system was also upgraded with OpenMPI to implement the open source Message Passing Interface protocol.

The Sandia lab reported that this new software stack environment allows for more memory per node for parallel jobs at runtime, as well as an increase in reliability and scalability of jobs.

Sandia researchers use Thunderbird to perform a broad range of weapons simulations, including atomistic scale-to-device modelling of radiation effects on semiconductor electronics, assessing weapon response safety in extreme thermal and impact environments, and quantifying uncertainties in weapons performance.

The Top 500 ranking of supercomputers is based on the Linpack benchmark.