350 likes | 491 Views
Outline. PRISM: goals & benefits FP5 project and the Support Initiative organisation the PRISM Areas of Expertise OASIS: historical background the community today some key notes The OASIS4 coupler: model adaptation component model description coupled model configuration
E N D
Outline • PRISM: • goals & benefits • FP5 project and the Support Initiative • organisation • the PRISM Areas of Expertise • OASIS: • historical background • the community today • some key notes • The OASIS4 coupler: • model adaptation • component model description • coupled model configuration • communication • regridding/transformations • grids supported • future developments and conclusions
PRISM: the goals Metadata Source management tomorrow Compiling + running env. Coupling & I/O Science MPI – LAPACK - … today tomorrow Fortran & C Compilers • Increase what Earth system modellers have in common today (compilers, message passing libraries, algebra libraries, etc.) • Share the development, maintenance and support of a wider set of Earth System Modelling software tools and standards.
PRISM: the benefits • reduce the technical development efforts of each individual research team • facilitate assembling, running, monitoring, and post-processing of ESMs based on state-of-the-art component models Help climate modellers spend more time on science: • promote the key scientific diversity • increase scientific collaboration • stimulate computer manufacturer contribution (tool portability, optimization of next generation of platforms for ESM needs)
PRISM: FP5 project and Support Initiative http://prism.enes.org • 2001-2004: the PRISM EU project • a European project funded for 4.8 M€ by the EC • 22 partners • 2005-2008: the PRISM Support Initiative: • 7 partners: CERFACS, CGAM, CNRS, ECMWF, MPI-M&D, UK MetOffice, NEC-CCRLE • 9 associate partners: CSC, CRAY, IPSL, Météo-France, MPI-M, NEC-HPCE, SGI, SMHI, SUN • 8 py/y for 3 years
PRISM: the organisation PRISM SB Chair P. Bougeault (ECMWF) PRISM Steering Board CERFACS, CGAM, CNRS, ECMWF, MPI-M&D, UK MetOffice, NEC-CCRLE PRISM Coordinator(s) E. Guilyardi (CNRS), S. Valcke (CERFACS) PUG chair R. Budich (MPI) PRISM Core Group (7 people) PRISM User Group PRISM areas of expertise (PAEs) Code Coupling & I/O Integration & modelling environments Data processing, visualisation and management Meta data Computing
PRISM: the Areas of Expertise PRISM is organised around 5 “PRISM Areas of Expertise”: • Promotion and, if needed, development of software tools for ESM • Organisation of related network of experts • Technology watch • Promotion of community standards • Coordination with other international efforts
PAE Code Coupling and IO Leader: S. Valcke (CERFACS) • development and support of tools for coupling climate modelling component codes: • OASIS3 and OASIS4 couplers • technology watch on coupling tools developed outside PRISM: • PALM coupler (CERFACS) • Bespoke Framework Generator (U. of Manchester) • CCSM (NCAR), … • relations with projects involving code coupling: • UK Met Office FLUME project • US ESMF project, GENIE project • ACCESS
PAE Integration and modelling environments Leader: M. Carter (UK MetOffice) • source version control for software development • code extraction and compilation • job configuration & running • Subversion for source version control (PRISM SVN server in Hamburg) • Standard Compiling and Running Environments (SCE/SRE, MPI-M&D): • integrated environment to compile and run different coupled models based on different component models (standard directory structure, coding rules) • Flexible Configuration Management (FCM, UK Met Office): • version control, compilation • prepIFS (ECMWF): • tailored graphical interface for model configuration (Web services tech.) • prepOASIS4: GUI to configure a coupled model using OASIS4 • Supervisor Monitor Scheduler (SMS, ECMWF): • management of networks of jobs across a number of platforms
PAE Data processing, visualisation and management Leader: M. Lautenschlager (MPI-M&D) • data processing, visualisation, archiving and exchange for Earth system research • standards and infrastructure for networking between geographically distributed archives • CDO (Climate data Operators form MPI-M) • CDAT (Climate Data Analysis Tools, from PCMDI) • CERA-2 data model (World Climate Data Centre): • geo-referenced climate data detection, browse and use • MARS (ECMWF): • meteorological data access and manipulation (for major NWP sites)
PAE Meta-data Leader: L. Steenman-Clark (CGAM) Meta-data: data about data, models, runs, ... … a hot topic in the last few years • exchange and use of data • interchangeability of Earth system models or modelling components • forum to discuss, develop, and coordinate metadata issues: • Numerical Model Metadata (U. of Reading): numerical code bases, simulations • CURATOR project (USA) : data, codes, simulations • Numerical grid metadata (GFDL, USA): grid • netCDF CF convention (PCMDI and BADC): climate and forecast data files • OASIS4 metadata coupling and IO interface • UK Met Office FLUME project: management of model configuration
PAE Computing Leader: M.-A. Foujols (IPSL), R. Redler (NEC-CCRLE) • Computing aspects are highly important for Earth system modelling • Computer vendors have to be kept informed about requirements emerging from the climate modelling community • Earth system modellers have to be informed about computing issues to preview difficulties and evolutions • file IO and data storage • algorithmic development • portable software to fit the needs of parallel and vector systems • sharing of experience (e.g. work on the Earth Simulator) • establishment of links with computing projects (e.g. DEISA) • information about important conferences and workshops.
OASIS: historical background OASIS: developed since 1991 to couple existing GCMs 19912001 |-- |--- PRISM OASIS 1 OASIS 2 OASIS3 OASIS4 OASIS1, OASIS2, OASIS3: • low resolution, low number of 2D fields, low coupling frequency: • flexibility very important, efficiency not so much! OASIS4: • higher resolution parallel models, massively parallelplatforms, 3D fields • need to optimise and parallelise the coupler
OASIS: the community today • CERFACS (France) ARPEGE3 - ORCA2-LIM ARPEGE4 - NEMO-LIM - TRIP • METEO-FRANCE (France)ARPEGE4 - ORCA2 ARPEGE medias - OPAmed ARPEGE3 - OPA8.1-GELATO • IPSL- LODYC, LMD, LSCE (France) LMDz - ORCA2LIM LMDz - ORCA4 • MERCATOR (France) (for interpolation only) • MPI - M&D (Germany) ECHAM5 - MPI-OM ECHAM5 - C-HOPE PUMA - C-HOPE EMAD - E-HOPE ECHAM5 - E-HOPE ECHAM4 - E-HOPE • ECMWFIFS - CTMIFS - ORCA2
OASIS: the community today • IFM-GEOMAR (Germany) ECHAM5 - NEMO (OPA9-LIM) • NCAS / U. Reading (UK) ECHAM4 - ORCA2 HADAM3-ORCA2 • SMHI (Sweden) RCA(region.) – RCO(region.) • NERSC (Norway) ARPEGE - MICOM • KNMI (Netherlands) ECHAM5 - TM5/MPI-OM • INGV (Italy) ECHAM5 – MPI-OM • ENEA (Italy) MITgcm - REGgcm • JAMSTEC (Japan) ECHAM5(T106) - ORCA ½ deg • IAP-CAS (China) AGCM - LSM • BMRC (Australia) TCLAPS - MOM • U. of Tasmania (Australia) Sea Ice code - MOM4 • RPN-Environment Canada (Canada) MEC - GOM • UQAM (Canada) GEM - RCO • U. Mississippi (USA) MM5 - HYCOM • IRI (USA) ECHAM5 - MOM3 • JPL (USA) UCLA-QTCM - Trident-Ind4-Atlantic
OASIS: Some key notes • Developers: CERFACS, NEC CCRLE, CNRS, SGI, (NEC HPCE) • Public domain; open source license (LGPL) • Programming language: Fortran 90 and C • Public domain libraries; vendor optimized versions may exist: • MPI1 and/or MPI2; NetCDF/parallel NetCDF; libXML • mpp_io; SCRIP • Static coupling
The OASIS3 coupler • Coupler developed since more than 15 years in CERFACS • Stable, well-debugged, but limited • Last version: oasis3_prism_2-5 delivered in September 2006 • User support provided but most development efforts go to OASIS4 • Mono-process coupler + parallel coupling library (PSMILe) • synchronisation of the component models • coupling fields exchange • I/O actions • mono-process interpolation • Platforms: • Fujitsu VPP5000, NEC SX5-6-8, Linux PC, IBM Power4, CRAY XD1, Compaq, SGI Origin, SGI O3400
OASIS3: model adaptation PRISM System Model Interface Library (PSMILe) API : • Initialization • Global grid definition (master process only) • Local partition definition • Coupling or I/O field declaration • Coupling or I/O field sending and receiving: • in model time stepping loop • depending on user’s specifications in namcouple: • user’s defined source or target (end-point communication) • coupling or I/O sending or receiving at appropriate times • automatic averaging/accumulation • automatic writing of coupling restart file at end of run • call prism_put (var_id, time, var_array, ierr) • call prism_get (var_id, time, var_array, ierr)
OASIS3: coupled model configuration • In text file namcouple, read by OASIS3 main process, and distributed to component model PSMILes at beginning of run: • total run time • component models • number of coupling fields • for each coupling field: • source and target names (end-point communication) (var_name) • grid acronym (grid_name) • coupling and/or I/O status • coupling or I/O period • transformations/interpolations
OASIS3: communication O A O • Parallel communication between parallel models and Oasis3 interpolation process A Oasis3 O A O B A B A • Direct communication between models with same grid and partitioning B A • I/O functionality (automatic switch between coupled and forced mode): GFDL mpp_io library A A file A PSMILe based on MPI1 or MPI2 message passing
OASIS3: interpolations/transformations O Oasis3 A O A Oasis3 O A O • separate sequential process • neighbourhood search • weight calculation • interpolation per se during the run • on 2D scalar or vector fields • Interfacing with RPN Fast Scalar INTerpolator package • nearest-neighbour, bilinear, bicubic for regular Lat-Lon grids • Interfacing with SCRIP1.4 library • nearest-neighbour, 1st and 2nd order conservative remapping for all grids • bilinear and bicubic interpolation for «logically-rectangular» grids • Bilinear and bicubic interpolation for reduced atmospheric grids • Other spatial transformations: flux correction, merging, etc. • General algebraic operations
The OASIS4 coupler • A parallel central Driver/Transformer: • launches models at the beginning of the run (MPI2) • reads the user-defined configuration information and distributes it to the component PSMILes • performs parallel transformations of the coupling fields during the run • A parallel model interface library (PSMILe) that performs: • weight-and-address calculation for the coupling field interpolations • MPI-based coupling exchanges between components • component I/O (GFDL mpp_io library)
OASIS4: model adaptation(1/3) • Initialization: call prism_init_comp(comp_id, comp_name, ierr) call prism_get_localcomm(comp_id, local_comm, ierr) • Definition of grid (3D) call prism_def_grid(grd_id, grd_name, comp_id, grd_shape, type, ierr) call prism_set_corners(grd_id, nbr_crnr, crnr_shape, crnr_array, ierr) • Placement of scalar points and mask on the grid: call prism_set_points (pt_id, pt_name, grd_id, pt_shape, pt_lon, pt_lat, pt_vert ,ierr) call prism_set_mask(msk_id, grd_id, msk_shape, msk_array, ierr) • Function overloading to keep the interface concise and flexible
OASIS4: model adaptation(2/3) prism_def_grid Example prism-set-corners prism_set_points prism_set_points prism_set_points prism_set_mask
OASIS4: model adaptation (3/3) • Coupling or I/O field declaration call prism_def_var (var_id, var_name, grd_id, pt_id, msk_id, var_nbrdims, var_shape, var_type, ierr) • End of definition call prism_enddef(ierr) • Coupling or I/O field sending and receiving: • in model time stepping loop • depending on user’s specifications in SMIOC: • user’s defined source or target, component or file (end-point communication) • coupling or I/O sending or receiving at appropriate times • averaging/accumulation call prism_put (var_id, date, date_bounds, var_array, info, ierr) call prism_get (var_id, date, date_bounds, var_array, info, ierr) • Termination call prism_terminate(ierr)
OASIS4: component model description Application and component description (XML files): • For each application (code): one Application Description (AD): • possible number of processes • components included • For each component in the application: one Potential Model Input and Output Description (PMIOD) • component general characteristics: name, component simulated, … • grid information: domain, resolution(s), grid type, … • potential I/O or coupling variables: • local name, standard name • units, valid min and max • numerical type • associated grid and points • intent –input and/or output
OASIS4: coupled model configuration (Through a GUI), the user produces • a Specific Coupling Configuration (SCC): • experiment and run start date and end date • applications, components for each application • host(s), number of processes per host, ranks for each component • For each component, a Specific Model Input and Output Configuration (SMIOC) • grid information: chosen resolution, … • I/O or coupling variables: • name, units, valid min max, numerical type, grid • activated intent –input and/or output • source and/or target (component and/or file) • coupling or I/O dates • transformations/interpolations/combinations
ATM-LAND AD Driver ATM PMIOD Va1: in, metadata Va1 Va2: out, metadata Va2 Va3: out, metadata Va3 user user OCE AD user OCE PMIOD Vo1: out, metadata Vo1 Vo2: in, metadata Vo2 user • SCC • ATM:... • OCE:... • LAND:... OCE OCE SMIOC Vo1 : to ATM Va1, T1 to File1 Vo2 : from ATM Vo1 ATM ATM SMIOC Va1 : from OCE Vo1 Va2: to OCE Vo2, T2 Va3 : to LAND Vl1 T LAND Va2 Vo2 Va3 Vl3 Va1 Vo1 LAND PMIOD Vl1: in, metadata Vl1 Vl2: in, metadata Vl2 user Vl2 File1 File2 Definition Phase LAND SMIOC Vl1 : from ATM Va3 Vl2 : from file File2 Composition Phase Deployment Phase
OASIS4 communication(1/2) OB C OB T C O2 O1 C C OB OB C C T O2 O1 C C OB OB C C Different grid and decomposition • via parallel Transformer: • for each target point, parallel calculation of source interpolation neighbours in source PSMILe Same grid, different decomposition • direct repartitioning: • for each target point, parallel calculation of source matching point in source PSMILe • Model interface library: PSMILe based on MPI1 or MPI2 • Parallel communication including repartitioning: • based on geographical description of the partitions • parallel calculation of communication patterns in source PSMILe
A file1 A A A A file file2 file A A A file3 A single distributed parallel OASIS4: communication(2/2) • end-point communication (source doesn’t know target and vice-versa) • parallel 3D neighbourhood search, based on efficient multigrid algorithm, in each source process PSMILe • extraction of useful part of source field only • one-to-one, one-to-many • parallel I/O (vector, bundles, vector bundles): GFDL mpp_io, parNetCDF
OASIS4 regridding/transformations • source time transformations (prism_put): • average, accumulation • target time transformations (prism_get) • time interpolation (for I/O only) • statistics • local transformations: • addition/multiplication by scalar • interpolation/regridding (3D): • nearest-neighbour 2D in the horizontal, “none” in the vertical • nearest-neighbour 3D • bilinear in the horizontal, “none” in the vertical • bicubic (gradient, 16 nghbrs) in the horizontal, “none” in the vertical • trilinear
Grids supported by OASIS4 • Regridding, repartitioning, I/O: • Regular in lon, lat, vert (“Reglonlatvrt”): • lon(i), lat(j), height(k) • Irregular in lon and lat, regular in the vert (“irrlonlat_regvrt”): • lon(i,j), lat(i,j), height(k) • Irregular in lon, lat, and vert (“irrlonlatvrt”) (not fully tested) • lon(i,j,k), lat(i,j,k), height(i,j,k) • Gaussian Reduced in lon and lat, regular in the vert (“Gaussreduced_regvrt”) • lon(nbr_pt_hor), lat(nbr_pt_hor), height(k) • Repartitioning and I/O only: • “Non-geographical” fields • no geographical information attached • local partitions described in the global index space (prism_def_partition) • I/O only: • Unstructured grids (“unstructlonlatvrt”) • lon(npt_tot), lat(npt_tot), height(npt_tot)
OASIS4: developments & perspectives • OASIS4 tested and run with toy examples on: • NEC SX6 and SX8 • IBM Power4 • SGI O3000/2000 • SGI IA64 Linux server Altix 3000/4000 • Intel® Xeon™ Infiniband and Myrinet Clusters • Linux PC DELL • OASIS4 now being used in a reduced number of real applications: • EU project GEMS: atmospheric dynamic and chemistry coupling • SMHI: ocean-atmosphere regional coupling • UK Met Office: global ocean-atmosphere coupling (currently prototyping) • IFM-GEOMAR (Kiel) in pseudo-models to interpolate high-resolution fields. • Current developments: • 2D conservative remapping • Parallel global search for the interpolation • Transformer efficiency • Full validation of current transformations • Full public release planned beginning 2007.
Conclusions PRISM provides: • framework promoting common software tools for Earth system modelling • some (more or less) standard tools (OASIS, source management, compiling, …) • network allowing ESM developers to share expertise and ideas • visible entry point for international coordination • metadata definition • WCRP white paper with ESMF on “Common Modeling Infrastructure for the International Community” • PRISM current decentralized organisation (bottom-up approach): • allows “best of breed” tools to naturally emerge • relies on the developments done in the different partner groups • Additional funding needed: • more networking and coordination activities • specific technical developments within the partner groups • Additional contributors are most welcome to join !