You are here: Home » Overview » Partners » BSC

BSC

Barcelona Supercomputing Center

The Barcelona Supercomputing Center-Centro Nacional de Supercomputación (BSC-CNS, BSC henceforth), created in 2005, has the mission to research, develop and manage information technology in order to facilitate scientific progress not only in computer science but also in a large range of applications. More than 350 people from 40 different countries perform and facilitate research into Computer Sciences, Life Sciences, Earth Sciences and Computational Applications in Science and Engineering at the BSC. The BSC is one of the eight Spanish “Severo Ochoa Centre of Excellence” institutions selected in the first round of this prestigious programme in Spain, as well as one of the four hosting members of the European PRACE Research Infrastructure. The BSC hosts MareNostrum III, a Tier-0 PRACE system currently ranked as the 24th most powerful supercomputer in Europe (57th in the world) with 1 Pflop/s capacity. In addition, the BSC hosts other High-Performance Computing (HPC) resources, among which it is worth mentioning MinoTauro, a hybrid system with GPUs incorporated.

The Earth Sciences Department of the BSC (ES-BSC) was established with the objective of carrying out research in Earth system modelling. The ES-BSC conducts research on air quality, mineral dust and climate modelling and strongly contributes to the scientific and technological advancement in atmospheric and mineral dust modelling. In this sense, the ES-BSC develops and maintains a state-of-the-art mineral dust model: NMMB/BSC-Dust. The excellent results of the group on this field have contributed to the recently creation of the first World Meteorological Organization (WMO) Regional Meteorological Center specialized on Atmospheric Sand and Dust Forecast, the “Barcelona Dust Forecast Center”. In which the NMMB/BSC-Dust model has been selected as the reference mineral dust model. Currently the model provides mineral dust forecasts to the World Meteorological Organization (WMO) Sand and Dust Storm Warning Advisory and Assessment System (SDS-WAS) Northern Africa- Middle East-Europe (NAMEE) Regional Centre that is managed by a consortium between the Spanish Weather Service (AEMET) and BSC. Furthermore, BSC and UNEP are collaborating in the development and implementation of the WMO SDS-WAS West Asia Regional Centre in which the NMMB/BSC-Dust is designed to perform mineral dust forecast simulations. ES-BSC also undertakes research on the development and assessment of dynamical and statistical methods for the prediction of global and regional climate on time scales ranging from a few weeks to several years. The EC-Earth model is used for this purpose. The formulation of the predictions includes the development and implementation of techniques to statistically downscale, calibrate and combine dynamical ensemble and empirical forecasts to satisfy specific user needs in the framework of the development of a climate service.

The high performance capabilities of MareNostrum III and the close collaboration with the Computer Sciences department allow efficiently increasing the spatial and temporal resolution of atmospheric modelling systems, in order to improve our knowledge on dynamic patterns of air pollutants in complex terrains and interactions and feedbacks of physico-chemical processes occurring in the atmosphere, as well as to push the boundaries of global climate prediction. To this, it should be added the increasing collaboration between the different BSC departments on the rapidly growing field of Big Data. Therefore, BSC offers a unique infrastructure to carry out the range of Earth system simulations on which the ES-BSC is a reference worldwide.

Role in the project

BSC’s contribution to ESiWACE is concentrated in WPs 1, 2 and 3. The centre has a vast experience in usability issue as the group not only uses the BSC infrastructure but also relies on several other machines. The department has managed to deploy climate models in platforms distributed around the globe like Archer (Edinburgh), Jaguar (USA), Lindgren (Sweden) and Curie (France). BSC has already organized a summer school on climate modelling (2nd E2SCMS) and therefore has the knowledge required to successfully performing the task. We will put our expertise coleading WP3 with MPI-M.

Due to the nature of the institution which is partially devoted to computer science, we also bring expertise to the WP2 (Scalability) leading task 2.3. ES-BSC can offer a wide range of services (based on software and hardware) to ESiWACE to analyse and improve performance of Earth system models. Increasing climate model efficiency requires the application of a wide set of tools to analyse and understand the behaviour of these models running in a parallel environment. BSC, through the Tools Group, develops tools like Paraver or Dimemas that can easily help the user of these codes to understand the behaviour of the code and identify possible bottlenecks and hardware related problems of the application. Furthermore, BSC provides the OmpSs programming model and COMPSs Superscalar (both developed in the institution). OmpSs can exploit parallelism through data dependencies or use different devices (GPU’s, accelerators) in a transparent way for the user and COMPSs Superscalar is a programming model that aims to ease the development of applications for distributed infrastructures, such as clusters, grids and clouds.

Relevant infrastructure and services available for climate & weather

BSC-CNS hosts MareNostrum, the most powerful supercomputer in Spain. MareNostrum has a peak performance of 1,1 Petaflops, with 48,896 Intel Sandy Bridge processors in 3,056 nodes, and 84 Xeon Phi 5110P in 42 nodes, with more than 104.6 TB of main memory and 2 PB of GPFS disk storage. This infrastructure will be upgraded during 2016.

BSC-CNS hosts MinoTauro, an NVIDIA GPU cluster with 64 nodes, each one carrying two Intel E5649 (6-Core) processor and two M2090 NVIDIA GPU cards. The system will be extended with 39 more nodes with Intel Haswell processors and 2 nvidia K80 each.

BSC-CNS has a dedicated storage for the Earth Sciences Department of more than 600 TB net space that will be grown to 2 PB in the following two years.

BSC-CNS will also deploy in the coming years a ESGF node.

Website

http://www.bsc.es

Wiki for the Earth Science department at https://earth.bsc.es/wiki/doku.php

Code repository at https://earth.bsc.es/gitlab/

© Copyright ESiWACE 2015