Porting and performance analysis of EC-Earth System on the BSC-MareNostrum Supercomputer

Earth System Models (ESM) have been successfuly developed over past few years, and are currently beeing used for simulating present day-climate, seasonal to interanual predictions of climate change. The supercomputer performance plays an important role in climate modeling since one of the challenging issues for climate modellers is to efficiently and accurately couple earth System components on present day computers architectures. At the Barcelona Supercomputing Center (BSC), we work with the EC- Earth System Model. The EC- Earth is an ESM, which currently consists of an atmosphere (IFS) and an ocean (NEMO) model that communicate with each other through the OASIS coupler. Additional modules (e.g. for chemistry and vegetation ) are under development. The EC-Earth ESM has been ported successfully over diferent high performance computin platforms (e.g, IBM P6 AIX, CRAY XT-5, Intelbased Linux Clusters, SGI Altix) at diferent sites in Europ (e.g., KNMI, ICHEC, ECMWF). The objective of the first phase of the project was to identify and document the issues related with the portability and performance of EC-Earth on the MareNostrum supercomputer, a System based on IBM PowerPC 970MP processors and run under a Linux Suse Distribution. EC-Earth was successfully ported to MareNostrum, and a compilation incompatibilty was solved by a two step compilation approach using XLF version 10.1 and 12.1 compilers. In addition, the EC-Earth performance was analyzed with respect to escalability and trace analysis with the Paravear software. This analysis showed that EC-Earth with a larger number of IFS CPUs (<128) is not feasible at the moment since some issues exists with the IFS-NEMO balance and MPI Communications.

Symmetry insights for design of supercomputer network topologies: roots and weights lattices

We present a family of networks whose local interconnection topologies are generated by the root vectors of a semi-simple complex Lie algebra. Cartan classification theorem of those algebras ensures those families of interconnection topologies to be exhaustive. The global arrangement of the network is defined in terms of integer or half-integer weight lattices. The mesh or torus topologies that network millions of processing cores, such as those in the IBM BlueGene series, are the simplest member of that category. The symmetries of the root systems of an algebra, manifested by their Weyl group, lends great convenience for the design and analysis of hardware architecture, algorithms and programs.

Cluster Supercomputer Architecture

The paper describes the architecture of SCIT - supercomputer system of cluster type and the base architecture features used during this research project. This supercomputer system is put into research operation in Glushkov Institute of Cybernetics NAS of Ukraine from the early 2006 year. The paper may be useful for those scientists and engineers that are practically engaged in a cluster supercomputer systems design, integration and services.

SCIT — Ukrainian Supercomputer Project

The paper describes a first supercomputer cluster project in Ukraine, its hardware, software and characteristics. The paper shows the performance results received on systems that were built. There are also shortly described software packages made by cluster users that have already made a return of investments into a cluster project.

Constraints on the infrared behavior of the ghost propagator in Yang-Mills theories

We present rigorous upper and lower bounds for the momentum-space ghost propagator G(p) of Yang-Mills theories in terms of the smallest nonzero eigenvalue (and of the corresponding eigenvector) of the Faddeev-Popov matrix. We apply our analysis to data from simulations of SU(2) lattice gauge theory in Landau gauge, using the largest lattice sizes to date. Our results suggest that, in three and in four space-time dimensions, the Landau gauge ghost propagator is not enhanced as compared to its tree-level behavior. This is also seen in plots and fits of the ghost dressing function. In the two-dimensional case, on the other hand, we find that G(p) diverges as p(-2-2 kappa) with kappa approximate to 0.15, in agreement with A. Maas, Phys. Rev. D 75, 116004 (2007). We note that our discussion is general, although we make an application only to pure gauge theory in Landau gauge. Our simulations have been performed on the IBM supercomputer at the University of Sao Paulo.

Constraints on the infrared behavior of the gluon propagator in Yang-Mills theories

We present rigorous upper and lower bounds for the zero-momentum gluon propagator D(0) of Yang-Mills theories in terms of the average value of the gluon field. This allows us to perform a controlled extrapolation of lattice data to infinite volume, showing that the infrared limit of the Landau-gauge gluon propagator in SU(2) gauge theory is finite and nonzero in three and in four space-time dimensions. In the two-dimensional case, we find D(0)=0, in agreement with Maas. We suggest an explanation for these results. We note that our discussion is general, although we apply our analysis only to pure gauge theory in the Landau gauge. Simulations have been performed on the IBM supercomputer at the University of Sao Paulo.

A Study on Unconditionally Stable FDTD Methods for the Modeling of Metamaterials

We assess the performance of three unconditionally stable finite-difference time-domain (FDTD) methods for the modeling of doubly dispersive metamaterials: 1) locally one-dimensional FDTD; 2) locally one-dimensional FDTD with Strang splitting; and (3) alternating direction implicit FDTD. We use both double-negative media and zero-index media as benchmarks.

Numerical Study of Hypersonic Leeward Flow over a Blunt Nosed Delta Wing

A kinetic theory based Navier-Stokes solver has been implemented on a parallel supercomputer (Intel iPSC Touchstone Delta) to study the leeward flowfield of a blunt nosed delta wing at 30-deg incidence at hypersonic speeds (similar to the proposed HERMES aerospace plane). Computational results are presented for a series of grids for both inviscid and laminar viscous flows at Reynolds numbers of 225,000 and 2.25 million. In addition, comparisons are made between the present and two independent calculations of the some flows (by L. LeToullec and P. Guillen, and S. Menne) which were presented at the Workshop on Hypersonic Flows for Re-entry Problems, Antibes, France, 1991.

A numerical study of large sparse matrix exponentials arising in Markov chains

Krylov subspace techniques have been shown to yield robust methods for the numerical computation of large sparse matrix exponentials and especially the transient solutions of Markov Chains. The attractiveness of these methods results from the fact that they allow us to compute the action of a matrix exponential operator on an operand vector without having to compute, explicitly, the matrix exponential in isolation. In this paper we compare a Krylov-based method with some of the current approaches used for computing transient solutions of Markov chains. After a brief synthesis of the features of the methods used, wide-ranging numerical comparisons are performed on a power challenge array supercomputer on three different models. (C) 1999 Elsevier Science B.V. All rights reserved.AMS Classification: 65F99; 65L05; 65U05.

A Grid Supercomputing Enviroment for High Demand Computational Applications

Despite the huge increase in processor and interprocessor network performace, many computational problems remain unsolved due to lack of some critical resources such as floating point sustained performance, memory bandwidth, etc... Examples of these problems are found in areas of climate research, biology, astrophysics, high energy physics (montecarlo simulations) and artificial intelligence, among others. For some of these problems, computing resources of a single supercomputing facility can be 1 or 2 orders of magnitude apart from the resources needed to solve some them. Supercomputer centers have to face an increasing demand on processing performance, with the direct consequence of an increasing number of processors and systems, resulting in a more difficult administration of HPC resources and the need for more physical space, higher electrical power consumption and improved air conditioning, among other problems. Some of the previous problems can´t be easily solved, so grid computing, intended as a technology enabling the addition and consolidation of computing power, can help in solving large scale supercomputing problems. In this document, we describe how 2 supercomputing facilities in Spain joined their resources to solve a problem of this kind. The objectives of this experience were, among others, to demonstrate that such a cooperation can enable the solution of bigger dimension problems and to measure the efficiency that could be achieved. In this document we show some preliminary results of this experience and to what extend these objectives were achieved.

Desenvolupament de codis numèrics per a la resolució de fluxos turbulents en ordinadors paral·lels emprant discretitzacions tipus symmetry-preserving

Projecte de recerca elaborat a partir d’una estada a la University of Groningen, Holanda, entre 2007 i 2009. La simulació directa de la turbulència (DNS) és una eina clau dins de la mecànica de fluids computacional. Per una banda permet conèixer millor la física de la turbulència i per l'altra els resultats obtinguts són claus per el desenvolupament dels models de turbulència. No obstant, el DNS no és una tècnica vàlida per a la gran majoria d'aplicacions industrials degut al elevats costos computacionals. Per tant, és necessari cert grau de modelització de la turbulència. En aquest context, s'han introduïts importants millores basades en la modelització del terme convectiu (no lineal) emprant symmetry-preserving regularizations. En tracta de modificar adequadament el terme convectiu a fi de reduir la producció d'escales més i més petites (vortex-stretching) tot mantenint tots els invariants de les equacions originals. Fins ara, aquest models s'han emprat amb èxit per nombres de Rayleigh (Ra) relativament elevats. En aquest punt, disposar de resultats DNS per a configuracions més complexes i nombres de Ra més elevats és clau. En aquest contexte, s'han dut a terme simulacions DNS en el supercomputador MareNostrum d'una Differentially Heated Cavity amb Ra=1e11 i Pr=0.71 durant el primer any dels dos que consta el projecte. A més a més, s'ha adaptat el codi a fi de poder simular el fluxe al voltant d'un cub sobre una pared amb Re=10000. Aquestes simulacions DNS són les més grans fetes fins ara per aquestes configuracions i la seva correcta modelització és un gran repte degut la complexitat dels fluxes. Aquestes noves simulacions DNS estan aportant nous coneixements a la física de la turbulència i aportant resultats indispensables per al progrés de les modelitzacións tipus symmetry-preserving regularization.

Ancient dispersal of the human fungal pathogen Cryptococcus gattii from the Amazon rainforest

Over the past two decades, several fungal outbreaks have occurred, including the high-profile 'Vancouver Island' and 'Pacific Northwest' outbreaks, caused by Cryptococcus gattii, which has affected hundreds of otherwise healthy humans and animals. Over the same time period, C. gattii was the cause of several additional case clusters at localities outside of the tropical and subtropical climate zones where the species normally occurs. In every case, the causative agent belongs to a previously rare genotype of C. gattii called AFLP6/VGII, but the origin of the outbreak clades remains enigmatic. Here we used phylogenetic and recombination analyses, based on AFLP and multiple MLST datasets, and coalescence gene genealogy to demonstrate that these outbreaks have arisen from a highly-recombining C. gattii population in the native rainforest of Northern Brazil. Thus the modern virulent C. gattii AFLP6/VGII outbreak lineages derived from mating events in South America and then dispersed to temperate regions where they cause serious infections in humans and animals.

Microcracks and Vortices in Superconducting Thin Films

In this work, superconducting YBa₂ Cu₃O_6+x (YBCO) thin films have been studied with the experimental focus on the anisotropy of BaZrO₃ (BZO) doped YBCOthin films and the theoretical focus on modelling flux pinning by numerically solving Ginzburg- Landau equations. Also, the structural properties of undoped YBCO thin films grown on NdGaO₃ (NGO) and MgO substrates were investigated. The thin film samples were made by pulsed laser ablation on single crystal substrates. The structural properties of the thin films were characterized by X-ray diffraction and atomic force microscope measurements. The superconducting properties were investigated with a magnetometer and also with transport measurements in pulsed magnetic field up to 30 T. Flux pinning was modelled by restricting the value of the order parameter inside the columnar pinning sites and then solving the Ginzburg-Landau equations numerically with the restrictions in place. The computations were done with a parallel code on a supercomputer. The YBCO thin films were seen to develop microcracks when grown on NGO or MgO substrates. The microcrack formation was connected to the structure of the YBCO thin films in both cases. Additionally, the microcracks can be avoided by careful optimization of the deposition parameters and the film thickness. The BZO doping of the YBCO thin films was seen to decrease the effective electron mass anisotropy, which was seen by fitting the Blatter scaling to the angle dependence of the upper critical field. The Ginzburg-Landau simulations were able to reproduce the measured magnetic field dependence of the critical current density for BZO doped and undoped YBCO. The simulations showed that in addition to the large density also the large size of the BZO nanorods is a key factor behind the change in the power law behaviour between BZO doped and undoped YBCO. Additionally, the Ginzburg-Landau equations were solved for type I thin films where giant vortices were seen to appear depending on the film thickness. The simulations predicted that singly quantized vortices are stable in type I films up to quite large thicknesses and that the size of the vortices increases with decreasing film thickness, in a way that is similar to the behaviour of the interaction length of Pearl vortices.