German English
Business news of the metropolitan region
HLRE-3_Stufe 1: © DKRZ

New Super Brain At Hamburg's German Climate Computing Centre

Mistral: like this wind, this new high performance computer comes from France. The French computer manufacturer Bull is currently installing the first phase of of this new system in Hamburg

“Mistral” will replace DKRZ’s IBM Power6 named “Blizzard” this summer. Blizzard has been serving climate research since 2009.

Peak Performance of 1.4 PFlops

Mistral, the new High Performance Computer System for Earth System Research (HLRE-3) consists of computer components by Bull, a disk storage system by Xyratex/Seagate and high performance network switches by Mellanox. These components are distributed over 41 racks weighing up to or even more than a ton, which are connected by bundles of fiber fabric.

About 1,500 compute nodes on the basis of Bullx B700 DLC systems with two 12-core Haswell processors each – a total of 36,000 cores – will reach a peak performance of 1.4 PFlops. This would be an increase by a factor of 9 compared with the Blizzard. At 75 TByte, the system’s main memory is 5 times larger than that of the IBM, and the new Lustre-based parallel file system will offer 20 PBytes, being more than three times larger than the current disk system. Furthermore, 12 visualization nodes equipped with NVidia Tesla K40 GPUs and 50 nodes for interactive access and pre- and post-processing of simulation data complement the system.

HLRE-3 Phase 1 Setup

Parallel to the installation works, DKRZ’s users have access to a small test system with 432 Haswell cores and a 300 TB Lustre file system for the purpose of preparing climate models for the new architecture.

The second phase of HLRE-3 is planned to be available in summer 2016. This extension will additionally roughly double computing and disk storage capacity. With a peak performance of 3 PFlops and a 50 PByte parallel file system, scientists can improve the regional resolution, account for more processes in the Earth system models or reduce uncertainties in climate projections.

source and further details:

More articles