CERFACS CERFACS Scientific Activity Report Jan. 2010 â Dec. 2011

More documents

Recommendations

Info

THE OASIS COUPLER 2.3 OASIS4 Since the beginning of the IS-ENES project in March 2009, CERFACS, CNRS and DKRZ (Deutsches Klimarechenzentrum GmbH, in Hamburg, Germany) have been collaborating on the development and evaluation of a new fully parallel version of the coupler, OASIS4. In particular, OASIS4 includes a library performing a fully parallel calculation of the source neighbour weights and addresses needed for the regridding of the coupling fields between the source and target grids (hereafter called the “neighbourhood search library”) originally developed by NEC Laboratories Europe - IT Research Division (NLE-IT). This work resulted in June 2011 in the release of the latest OASIS4 version, OASIS4 1beta ([CC29], [CC10]). OASIS4 was used by Météo-France, KNMI (Netherlands), and MPI-M (Germany) in the framework of the EU GEMS project (lead by ECMWF) for 3D coupling between atmospheric dynamic and atmospheric chemistry models and is currently used by SMHI (Sweden), the Alfred Wegener Institute for Polar and Marine Research (AWI in Germany), the BoM in Australia for ocean-atmosphere 2D regional or global coupling. In the framework of the METAFOR project, OASIS4 was adapted so to allow the use of the Common Information Model standard to configure the coupling exchanges ([CC28]). However, this work also lead us to conclude that OASIS4 parallel neighbourhood search library, originally developed by NEC, presents some fundamental weaknesses in its design. In particular, the support of unstructured grids was not originally included and it would be very difficult to add it in the current code. Also, it is now very clear that it was developed with efficiency as the prime criteria, leaving aside readability and ease of development. Therefore we decided to retrofit OASIS4 for pre-computed regridding weights, thereby circumventing its parallel neighbourhood search library, and use what we call the OASIS4 “user-defined” interpolation ([CC23]). To use this functionality, the user has to describe the links associating specific source grid points with specific target grid points in a separate NetCDF file. For each link, the index of the source point, the index of the target point and the weight associated to that link have to be provided. The OASIS4 communication library reads these indices and weights and automatically defines on each side a non-geographical grid with one point for each link. The multiplication of the source field values by the appropriate weights is done in parallel on the source side and a parallel redistribution of the results is done directly between the source and the target processes by the OASIS4 communication library. The userdefined regridding has been validated with simple toy models. It now has to be validated for real grids and in real coupled ESMs. 2.4 OASIS3-MCT The second solution is based on the modification of OASIS3 to use the Model Coupling Toolkit 5 (MCT) developed by the Argonne National Laboratory in the USA. MCT requires that the regridding weights are pre-computed offline, but then implements fully parallel regridding and exchanges of the coupling fields and has proven parallel performance. MCT embodies a generic approach for creating coupled applications. Its design philosophy, based on flexibility and minimal invasiveness, is close to the OASIS approach. MCT uses distributed objects to store the coupling data and pre-computed regridding weights as well as a domain decomposition descriptor (DDD) to describe the parallel decomposition of the component models. Parallel communication patterns are computed automatically from the source and target DDDs. To use MCT, one has to load the coupling data into MCT data types and call MCT parallel matrix multiplication (for regridding) and communication methods. Parallel data transfer is then accomplished by pairs of send/receive methods with coupling data and communication pattern as inputs. MCT is, most notably, the underlying coupling software used in National Center for Atmospheric Research Community Earth System Model 1 (NCAR CESM1) [1]. Interfacing MCT in OASIS3 will provide a valid and efficient solution to remove current OASIS3 bottleneck ; in this case, the coupling fields will be regridded in parallel in the source 5 http ://www.mcs.anl.gov/research/projects/mct/ 102 Jan. 2010 – Dec. 2011
COUPLING TOOLS AND HPC CLIMATE MODELING communication library and directly redistributed between the source and target parallel distribution without the need to gather the whole coupling field onto one OASIS3 process. One advantage of this solution is that it will be totally transparent for the OASIS3 user, in the sense that the current OASIS3 communication library API (Application Programming Interface) will not change. Technically, the two solutions described above, i.e. the OASIS4 user-defined regridding and the interfacing of MCT in OASIS3, are in principle equivalent. It will therefore be interesting to compare the two solutions and choose the most efficient one if they both prove to be valid options for massively parallel coupled simulations. 2.5 References [1] A. P. Craig, M. Vertenstein, and R. Jacob, (2011), A new flexible coupler for earth system modeling developed for CCSM4 and CESM1, International Journal of High Performance Computing Application, doi :10.1177/1094342011428141. [2] J.-L. Dufresne, M.-A. Foujols, A. Denvil, S. Caubel, O. Marti, O. Aumont, Y. Balkanski, S. Bekki, H. Bellenger, R. Benshila, S. Bony, L. Bopp, P. Braconnot, P. Brockmann, P. Cadule, F. Cheruy, F. Codron, A. Cozic, D. Cugnet, N. de Noblet, J.-P. Duvel, C. Ethé, L. Fairhead, T. Fichefet, S. Flavoni, P. Friedlingstein, J.-Y. Grandpeix, L. Guez, E. Guilyardi, D. Hauglustaine, F. Hourdin, A. Idelkadi, J. Ghattas, S. Joussaume, M. Kageyama, G. Krinner, S. Labetoulle, A. Lahellec, M.-P. Lefebvre, F. Lefevre, C. Levy, Z. X. Li, J. Lloyd, F. Lott, G. Madec, M. Mancip, M. Marchand, S. Masson, Y. Meurdesoif, J. Mignot, I. Musat, S. Parouty, J. Polcher, C. Rio, M. Schulz, D. Swingedouw, S. Szopa, C. Talandier, P. Terray, and Viovy, (2012), Climate change projections using the IPSL- CM5 Earth System Model : from CMIP3 to CMIP5, Clim. Dyn. submitted. [3] W. Hazeleger, X. Wang, C. Severijns, S. Stefanescu, R. Bintanja, A. Sterl, K. Wyser, T. Semmler, Y. S., and B. van den Hurk, (2011), EC-Earth V2.2 : description and validation of a new seamless earth system prediction model, Clim. Dyn, DOI : 10.1007/s00382–011–1228–5. [4] E. Scoccimarro, S. Gualdi, A. Bellucci, A. Sanna, P. G. Fogli, E. Manzini, M. Vichi, P. Oddo, and A. Navarra, (2011), Effects of Tropical Cyclones on Ocean Heat Transport in a High Resolution Coupled General Circulation Model, Journal of Climate, 24, 4368–4384. [5] M. Vichi, E. Manzini, P. G. Fogli, A. Alessandri, L. Patara, E. Scoccimarro, M. S., and N. A., (2011), Global and regional ocean carbon uptake and climate change : sensitivity to a substantial mitigation scenario, Clim. Dyn, 37, DOI : 10.1007/s00382–011–1079–0. CERFACS ACTIVITY REPORT 103
Page 1 and 2:
CERFACS Scientific Activity Report
Page 3 and 4:
Table des matières 1 Foreword xiii
Page 5 and 6:
9 Publications 40 9.1 Books . . . .
Page 7 and 8:
3 The OpenPALM coupler 104 3.1 Intr
Page 9 and 10:
Table des figures CERFACS chart as
Page 11 and 12:
TABLE DES FIGURES 6 Computational F
Page 13 and 14:
Foreword Welcome to the 2010-2011 S
Page 15 and 16:
CERFACS STRUCTURE CERFACS organizat
Page 17 and 18:
CERFACS STAFF NAME POSITION PERIOD
Page 19 and 20:
CERFACS STAFF SILVA GARZON Ph.D stu
Page 21 and 22:
CERFACS STAFF NAME POSITION PERIOD
Page 23:
CERFACS Wide-Interest Seminars Pete
Page 27 and 28:
1 Introduction 1.1 Introduction The
Page 29 and 30:
PARALLEL ALGORITHMS PROJECT success
Page 31 and 32:
PARALLEL ALGORITHMS PROJECT Visitor
Page 33 and 34:
3 List of Members of HiePACS HiePAC
Page 35 and 36:
4 Dense and Sparse Matrix Computati
Page 37 and 38:
PARALLEL ALGORITHMS PROJECT Working
Page 39 and 40:
5 Iterative Methods and Preconditio
Page 41 and 42:
PARALLEL ALGORITHMS PROJECT 5.5 Pre
Page 43 and 44:
PARALLEL ALGORITHMS PROJECT We are
Page 45 and 46:
PARALLEL ALGORITHMS PROJECT is to e
Page 47 and 48:
PARALLEL ALGORITHMS PROJECT automor
Page 49 and 50:
PARALLEL ALGORITHMS PROJECT 7.4 Sol
Page 51 and 52:
PARALLEL ALGORITHMS PROJECT precond
Page 53 and 54:
PARALLEL ALGORITHMS PROJECT 7.16 An
Page 55 and 56:
PARALLEL ALGORITHMS PROJECT solutio
Page 57 and 58:
8 Conferences and Seminars 8.1 Conf
Page 59 and 60:
PARALLEL ALGORITHMS PROJECT Decembe
Page 61 and 62:
PARALLEL ALGORITHMS PROJECT July WC
Page 63 and 64:
PARALLEL ALGORITHMS PROJECT Septemb
Page 65 and 66:
PARALLEL ALGORITHMS PROJECT [ALG15]
Page 67:
PARALLEL ALGORITHMS PROJECT [ALG54]
Page 71 and 72:
1 Overview presentation The main ex
Page 73 and 74:
ELECTROMAGNETISM AND ACOUSTICS TEAM
Page 75 and 76: ELECTROMAGNETISM AND ACOUSTICS TEAM
Page 93: ELECTROMAGNETISM AND ACOUSTICS TEAM
Page 97 and 98: 1 Introduction Data assimilation is
Page 99 and 100: DATA ASSIMILATION 2.3 Calibrating o
Page 101 and 102: 3 Data assimilation for atmospheric
Page 103 and 104: DATA ASSIMILATION BECM using or not
Page 105 and 106: DATA ASSIMILATION the potential vor
Page 107 and 108: DATA ASSIMILATION A two years simul
Page 109 and 110: DATA ASSIMILATION FIG. 4.1: Reducti
Page 111 and 112: DATA ASSIMILATION FIG. 4.3: April 2
Page 113 and 114: DATA ASSIMILATION by a surface mode
Page 115 and 116: 6 Data assimilation with EDF neutro
Page 117 and 118: DATA ASSIMILATION demonstrate that
Page 119 and 120: DATA ASSIMILATION [DA13] S. Massart
Page 121: 4 Coupling tools and HPC Climate Mo
Page 124 and 125: 2 The OASIS coupler The OASIS coupl
Page 128 and 129: 3 The OpenPALM coupler 3.1 Introduc
Page 130 and 131: THE OPENPALM COUPLER In the NitroEu
Page 132 and 133: 4 HPC Climate modelling 4.1 General
Page 134 and 135: 5 Publications 5.1 Conference Proce
Page 137: 5 Climate Variability and Global Ch
Page 140 and 141: 2 Climate variability and predictab
Page 142 and 143: CLIMATE VARIABILITY AND PREDICTABIL
Page 144 and 145: CLIMATE VARIABILITY AND PREDICTABIL
Page 146 and 147: CLIMATE CHANGE AND IMPACT STUDIES a
Page 148 and 149: CLIMATE CHANGE AND IMPACT STUDIES b
Page 150 and 151: PUBLICATIONS [CM15] Y. Kwon, C. Des
Page 153 and 154: 1 Introduction The CFD (Computation
Page 155 and 156: 2 Combustion The research activity
Page 157 and 158: COMPUTATIONAL FLUID DYNAMICS rate-o
Page 159 and 160: COMPUTATIONAL FLUID DYNAMICS FIG. 2
Page 161 and 162: COMPUTATIONAL FLUID DYNAMICS an iss
Page 163 and 164: COMPUTATIONAL FLUID DYNAMICS TU Ber
Page 169 and 170: COMPUTATIONAL FLUID DYNAMICS design
Page 171 and 172: COMPUTATIONAL FLUID DYNAMICS (a) (b
Page 173 and 174: COMPUTATIONAL FLUID DYNAMICS 3.1.5
Page 177 and 178:
COMPUTATIONAL FLUID DYNAMICS 3.2.5
Page 179 and 180:
COMPUTATIONAL FLUID DYNAMICS FIG. 3
Page 181 and 182:
COMPUTATIONAL FLUID DYNAMICS (a) Ve
Page 183 and 184:
COMPUTATIONAL FLUID DYNAMICS The cl
Page 185 and 186:
COMPUTATIONAL FLUID DYNAMICS FIG. 4
Page 187 and 188:
COMPUTATIONAL FLUID DYNAMICS (a) Do
Page 189 and 190:
COMPUTATIONAL FLUID DYNAMICS 4.3.3
Page 191 and 192:
5 Publications 5.1 Conferences Proc
Page 193 and 194:
COMPUTATIONAL FLUID DYNAMICS [CFD34
Page 195 and 196:
Page 197 and 198:
Page 199 and 200:
Page 201:
7 Aviation and Environment
Page 204 and 205:
INTRODUCTION instead of regular ker
Page 206 and 207:
SIMULATIONS OF AIRCRAFT WAKES AND R
Page 208 and 209:
Page 210 and 211:
Page 212 and 213:
LARGE SCALE ATMOSPHERIC COMPOSITION
Page 214 and 215:
LARGE SCALE ATMOSPHERIC COMPOSITION
Page 216:
PUBLICATIONS 4.2 Technical Reports
show all

CERFACS CERFACS Scientific Activity Report Jan. 2010 â Dec. 2011

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?

CERFACS CERFACS Scientific Activity Report Jan. 2010 â Dec. 2011