End-to-end optimization potentials in HPC applications for NWP and Climate Research
|
|
- Jean Russell
- 6 years ago
- Views:
Transcription
1 End-to-end optimization potentials in HPC applications for NWP and Climate Research Luis Kornblueh and Many Colleagues and DKRZ MAX-PLANCK-GESELLSCHAFT
2 ... or a guided tour through the jungle... MAX-PLANCK-GESELLSCHAFT
3 Joint DKRZ/MPIM initial brainstorming Courtesy of Joachim Biercamp, DKRZ Optimization potentials 3 / 25
4 Rationals Why do end-to-end management? Scientific responsible experimentation support Optimization potentials 4 / 25
5 Rationals Why do end-to-end management? Scientific responsible experimentation support Reduce workload of all members of the numerical experimentation community... Optimization potentials 4 / 25
6 An experimentation howto Title, statement of problem, and hypothesis Optimization potentials 5 / 25
7 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Optimization potentials 5 / 25
8 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Step-by-step description in such a way that the experiment can be repeated Optimization potentials 5 / 25
9 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Step-by-step description in such a way that the experiment can be repeated Run the experiment, NO, not three times only one time! Optimization potentials 5 / 25
10 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Step-by-step description in such a way that the experiment can be repeated Run the experiment, NO, not three times only one time! Observations made during and analysis of the results Optimization potentials 5 / 25
11 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Step-by-step description in such a way that the experiment can be repeated Run the experiment, NO, not three times only one time! Observations made during and analysis of the results Discuss possible errors that could have occurred in the collection of the data (experimental errors) Optimization potentials 5 / 25
12 An experimentation howto Title, statement of problem, and hypothesis Site, hardware, software stack, used programs (versioned), sources, and documented input and boundary data Step-by-step description in such a way that the experiment can be repeated Run the experiment, NO, not three times only one time! Observations made during and analysis of the results Discuss possible errors that could have occurred in the collection of the data (experimental errors) Publication of the results Optimization potentials 5 / 25
13 The context What is our context? complex, non-standardized workflows and toolchains various processing steps by various actors in climate research as well very often changing workflows Important to note - we are not doing mainstream computing! Optimization potentials 6 / 25
14 Target: provenance of the whole data life cycle Focus on adding: primary data generation primary data processing Already available to a large extent: data publishing secondary data processing data distribution further not controlable data processing Optimization potentials 7 / 25
15 Scientists define experiment easily should organize their individual experiments workflow easily should be enabled to program their individual tasks should not care on collecting provenance data Optimization potentials 8 / 25
16 Scientific programmers define experiment easily should organize experiment workflows easily should be enabled to program individual tasks should not care on collecting provenance data should be enabled to query provenance data for bug tracking, performance improvements,... Optimization potentials 9 / 25
17 Computing center staff should not care on collecting provenance data should be enabled to query provenance data for failure analysis, performance improvements,... Optimization potentials 10 / 25
18 Components of the basic systems Frameworks Setup GUI or CLI Domain Specific Language for Task Scripting (Provenance Collecting Enabled) Cylc Meta-scheduling Preparation Post-processing Model run Archiving Optimization potentials 11 / 25
19 Tools Packages in use: python as scripting language postgres - provenance data collection subversion/git (migration to git for parts or all later, if model developers get convinced) cmake (migration from autotools and self-maintained makefile generator) Web interface for site and compiler dependencies; dependencies versioned in line with model code namelist migration to xml as model source code, user interfaces: a kind of namelist and a GUI. Optimization potentials 12 / 25
20 Single task component Scripting user friendly modeling language restartability exception and error handling in scientist understandable form full support of modern programming concepts use the python hype to change from shell to python Optimization potentials 13 / 25
21 Model build requirements Standardized, convenient, and fast tools out-of-source build Optimization potentials 14 / 25
22 Model build requirements Standardized, convenient, and fast tools out-of-source build dependency resolution Optimization potentials 14 / 25
23 Model build requirements Standardized, convenient, and fast tools out-of-source build dependency resolution make, with hooks for actions at certain steps Optimization potentials 14 / 25
24 Model build requirements Standardized, convenient, and fast tools out-of-source build dependency resolution make, with hooks for actions at certain steps make test Optimization potentials 14 / 25
25 Model build requirements Standardized, convenient, and fast tools out-of-source build dependency resolution make, with hooks for actions at certain steps make test make install Optimization potentials 14 / 25
26 Use of optimized codes: mpiesm and icon In use vectorization (we never gave up on this!), hand gather, scatter, merge instead of standard conditionals, and exposing transcendentials nproma blocking for different architectures OpenMP orphaning - whole physics including radiation run in an single OpenMP directive MPI implementation constantly revisited including building up static load-balancing strategies real asynchronous parallel I/O invest in optimization of libraries Maybe give up on bit-reproducability for production, but not for development! Optimization potentials 15 / 25
27 Data compression On top compression for grib2: AEC (CCSDS algorithm). a standardization exercise got NASA US patent released reimplement from scratch to go around commercial copyright define grib2 template for WMO validate with independent software stacks (Q4 2014) Average reduction in data size over 4-byte float: factor 5, and for grib Encoding and decoding are really, really fast. Get this ported to netcdf4 (hdf5), started but process is slow. Joint work of Mathis Rosenhauer, DKRZ, Shahram Najm, ECMWF, Uwe Schulzweida, and Luis Kornblueh Optimization potentials 16 / 25
28 ICON ocean scaling improvements Wallclock time (optimized) Wallclock time (original) Wallclock time [s] MPI tasks Ocean scaling improvements: no land points, gmres restart, hybrid MPI/OpenMP, code rewrite. Courtesy of Leonidas Linardakis Optimization potentials 17 / 25
29 ECHAM6 scaling improvements Wallclock time [s] Wallclock (cdi) 10 Output (cdi) Wallclock (cdi-pio) 5 Output (cdi-pio) Output time [s] MPI tasks Change from serial (cdi) to parallel output (cdi-pio, using RDMA, I/O time on compute nodes is essentially zero). Courtesy of Irina Fast, Thomas Jahns, DKRZ, and Deike Kleberg Optimization potentials 18 / 25
30 Use of optimized post-processing tools Extend and improve our toolchain (cdi and cdo) Available basic optimized code compute intensive operators are OpenMP parallelized processing of data can be handled by an threaded pipelining method Future data streaming instead of file storage transfer DAG based processing for highest possible parallel efficiency database information system for online data Optimization potentials 19 / 25
31 Experiment organization cylc - the Meta-Scheduler design your own distributed suites of inter-dependent cycling tasks efficient, modular, and reusable validate and visualize workflows on the fly control your running suites diagnose failures (easily!?) simplify failure recovery benefit from expert experience with a specialized tool for meteorological forecasting systems Courtesy of Hilary Oliver, NIWA and contributors Optimization potentials 20 / 25
32 A task modeling framework (provenance data collection) Cylc controled tasks and provenance collection high level programming language python embedded abstraction layer for file operations abstract task description embedded provenance data collection (database stored) tightly connected to cylc connect provenance enabled workflow to ESGF data distribution Remark: introduces complexity reduction methods to all users Courtesy of Deike Kleberg, MPIM and Pavan Siligam, DKRZ Optimization potentials 21 / 25
33 Ensemble overview setup prepare sync_start models archiving sync postproc models ens_post postproc archiving sync ens_post ensemble Optimization potentials 22 / 25
34 Ensemble detail setup prep_1 sync_start prep_0 model_1 model_0 post_oce1 arch_atm1 arch_oce1 arch_lnd1 post_atm1 post_lnd1 sync post_atm0 post_oce0 post_lnd0 arch_lnd0 arch_atm0 arch_oce0 model_1 ens_post model_0 post_atm1 post_lnd1 arch_oce1 arch_atm1 arch_lnd1 post_oce1 sync post_oce0 post_atm0 arch_lnd0 arch_oce0 arch_atm0 post_lnd0 ens_post ensemble Optimization potentials 23 / 25
35 A cylc optimization step connect cylc from inside application (in C) use curl for submitting http/post request to a WebServer POST triggers CGI as interface to cylc Later: server part integrated into cylc Do not poll! It is expensive! Optimization potentials 24 / 25
36 What is coming next? Primary target: Get an optimized worflow system working AT OUR SITE! Keep being conservative in model source code adaptation: concentrate on what we have But: Outsource exploration of GPU handling (CSCS) and take over necessary changes Constantly observe development directions Explore CS concepts in the easier context of post-processing Do: Tutorials and training, training and tutorials, tutorials and training,... Optimization potentials 25 / 25
cdo Data Processing (and Production) Luis Kornblueh, Uwe Schulzweida, Deike Kleberg, Thomas Jahns, Irina Fast
cdo Data Processing (and Production) Luis Kornblueh, Uwe Schulzweida, Deike Kleberg, Thomas Jahns, Irina Fast Max-Planck-Institut für Meteorologie, DKRZ September 24, 2014 MAX-PLANCK-GESELLSCHAFT Data
More informationParallel I/O using standard data formats in climate and NWP
Parallel I/O using standard data formats in climate and NWP Project ScalES funded by BMBF Deike Kleberg Luis Kornblueh, Hamburg MAX-PLANCK-GESELLSCHAFT Outline Introduction Proposed Solution Implementation
More informationPorting and Optimizing the COSMOS coupled model on Power6
Porting and Optimizing the COSMOS coupled model on Power6 Luis Kornblueh Max Planck Institute for Meteorology November 5, 2008 L. Kornblueh, MPIM () echam5 November 5, 2008 1 / 21 Outline 1 Introduction
More informationThe Icosahedral Nonhydrostatic (ICON) Model
The Icosahedral Nonhydrostatic (ICON) Model Scalability on Massively Parallel Computer Architectures Florian Prill, DWD + the ICON team 15th ECMWF Workshop on HPC in Meteorology October 2, 2012 ICON =
More informationKepler Scientific Workflow and Climate Modeling
Kepler Scientific Workflow and Climate Modeling Ufuk Turuncoglu Istanbul Technical University Informatics Institute Cecelia DeLuca Sylvia Murphy NOAA/ESRL Computational Science and Engineering Dept. NESII
More informationDeutscher Wetterdienst. Ulrich Schättler Deutscher Wetterdienst Research and Development
Deutscher Wetterdienst COSMO, ICON and Computers Ulrich Schättler Deutscher Wetterdienst Research and Development Contents Problems of the COSMO-Model on HPC architectures POMPA and The ICON Model Outlook
More informationHPC Performance Advances for Existing US Navy NWP Systems
HPC Performance Advances for Existing US Navy NWP Systems Timothy Whitcomb, Kevin Viner Naval Research Laboratory Marine Meteorology Division Monterey, CA Matthew Turner DeVine Consulting, Monterey, CA
More informationUniform Resource Locator Wide Area Network World Climate Research Programme Coupled Model Intercomparison
Glossary API Application Programming Interface AR5 IPCC Assessment Report 4 ASCII American Standard Code for Information Interchange BUFR Binary Universal Form for the Representation of meteorological
More informationSoftware Infrastructure for Data Assimilation: Object Oriented Prediction System
Software Infrastructure for Data Assimilation: Object Oriented Prediction System Yannick Trémolet ECMWF Blueprints for Next-Generation Data Assimilation Systems, Boulder, March 2016 Why OOPS? Y. Trémolet
More informationAdvanced Parallel Programming. Is there life beyond MPI?
Advanced Parallel Programming Is there life beyond MPI? Outline MPI vs. High Level Languages Declarative Languages Map Reduce and Hadoop Shared Global Address Space Languages Charm++ ChaNGa ChaNGa on GPUs
More informationECMWF's Next Generation IO for the IFS Model and Product Generation
ECMWF's Next Generation IO for the IFS Model and Product Generation Future workflow adaptations Tiago Quintino, B. Raoult, S. Smart, A. Bonanni, F. Rathgeber, P. Bauer ECMWF tiago.quintino@ecmwf.int ECMWF
More informationJoachim Biercamp Deutsches Klimarechenzentrum (DKRZ) With input from Peter Bauer, Reinhard Budich, Sylvie Joussaume, Bryan Lawrence.
Joachim Biercamp Deutsches Klimarechenzentrum (DKRZ) With input from Peter Bauer, Reinhard Budich, Sylvie Joussaume, Bryan Lawrence. The ESiWACE project has received funding from the European Union s Horizon
More informationc y l c a high level introduction CSPP/IMAPP Users Group Meeting SSEC, Madison, Wisconsin, June
a high level introduction CSPP/IMAPP Users Group Meeting SSEC, Madison, Wisconsin, June 27 2017 Adapted from original slides by Hilary J Oliver, NIWA what s cylc? a workflow engine to construct complex,
More informationCDI-pio, CDI with parallel I/O
CDI-pio, CDI with parallel I/O Deike Kleberg, Uwe Schulzweida, Thomas Jahns and Luis Kornblueh Max-Planck-Institute for Meteorology and Deutsches Klima Rechenzentrum Project ScalES funded by BMBF February
More informationVisIt Libsim. An in-situ visualisation library
VisIt Libsim. An in-situ visualisation library December 2017 Jean M. Favre, CSCS Outline Motivations In-situ visualization In-situ processing strategies VisIt s libsim library Enable visualization in a
More informationDeutscher Wetterdienst
Porting Operational Models to Multi- and Many-Core Architectures Ulrich Schättler Deutscher Wetterdienst Oliver Fuhrer MeteoSchweiz Xavier Lapillonne MeteoSchweiz Contents Strong Scalability of the Operational
More informationParallel I/O Libraries and Techniques
Parallel I/O Libraries and Techniques Mark Howison User Services & Support I/O for scientifc data I/O is commonly used by scientific applications to: Store numerical output from simulations Load initial
More informationeccodes GRIB Fortran 90 - Python APIs Practicals 2 Dominique Lucas and Xavi Abellan
eccodes GRIB Fortran 90 - Python APIs Practicals 2 Dominique Lucas and Xavi Abellan Dominique.Lucas@ecmwf.int Xavier.Abellan@ecmwf.int ECMWF March 1, 2017 Practical 2: eccodes indexing ecgate$ cd $SCRATCH/eccodes_api_practicals/exercise2
More informationSoftware stack deployment for Earth System Modelling using Spack
Software stack deployment for Earth System Modelling using Spack Kim Serradell Maronda (BSC) Sergey Kosukhin (MPI-M) The ESiWACE project has received funding from the European Union s Horizon 2020 research
More informationWorking with Scientific Data in ArcGIS Platform
Working with Scientific Data in ArcGIS Platform Sudhir Raj Shrestha sshrestha@esri.com Hong Xu hxu@esri.com Esri User Conference, San Diego, CA. July 11, 2017 What we will cover today Scientific Multidimensional
More informationeccodes GRIB Fortran 90 - Python APIs Practicals 2 Dominique Lucas and Xavi Abellan
eccodes GRIB Fortran 90 - Python APIs Practicals 2 Dominique Lucas and Xavi Abellan Dominique.Lucas@ecmwf.int Xavier.Abellan@ecmwf.int ECMWF March 1, 2016 Practical 2: eccodes indexing ecgate$ cd $SCRATCH/eccodes_api_practicals/exercise2
More informationNew User Seminar: Part 2 (best practices)
New User Seminar: Part 2 (best practices) General Interest Seminar January 2015 Hugh Merz merz@sharcnet.ca Session Outline Submitting Jobs Minimizing queue waits Investigating jobs Checkpointing Efficiency
More informationIntroduction to Parallel Programming. Tuesday, April 17, 12
Introduction to Parallel Programming 1 Overview Parallel programming allows the user to use multiple cpus concurrently Reasons for parallel execution: shorten execution time by spreading the computational
More informationirods for Data Management and Archiving UGM 2018 Masilamani Subramanyam
irods for Data Management and Archiving UGM 2018 Masilamani Subramanyam Agenda Introduction Challenges Data Transfer Solution irods use in Data Transfer Solution irods Proof-of-Concept Q&A Introduction
More informationPrototyping an in-situ visualisation mini-app for the LFRic Project
Prototyping an in-situ visualisation mini-app for the LFRic Project Samantha V. Adams, Wolfgang Hayek 18th Workshop on high performance computing in meteorology 24 th -28 th September 2018, ECMWF, UK.
More informationData near processing support for climate data analysis. Stephan Kindermann, Carsten Ehbrecht Deutsches Klimarechenzentrum (DKRZ)
Data near processing support for climate data analysis Stephan Kindermann, Carsten Ehbrecht Deutsches Klimarechenzentrum (DKRZ) Overview Background / Motivation Climate community data infrastructure Data
More informationSSQA Seminar Series. Server Side Testing Frameworks. Sachin Bansal Sr. Quality Engineering Manager Adobe Systems Inc. February 13 th, 2007
SSQA Seminar Series Server Side Testing Frameworks Sachin Bansal Sr. Quality Engineering Manager Adobe Systems Inc. February 13 th, 2007 1 Agenda Introduction Drivers for Server Side Testing Challenges
More informationIndex Introduction Setting up an account Searching and accessing Download Advanced features
ESGF Earth System Grid Federation Tutorial Index Introduction Setting up an account Searching and accessing Download Advanced features Index Introduction IT Challenges of Climate Change Research ESGF Introduction
More informationDeutscher Wetterdienst
Accelerating Work at DWD Ulrich Schättler Deutscher Wetterdienst Roadmap Porting operational models: revisited Preparations for enabling practical work at DWD My first steps with the COSMO on a GPU First
More informationWELCOME to the PRACTICAL EXERCISES
WELCOME to the PRACTICAL EXERCISES 18.-22.02.2013 COSMO/CLM Training Course 2013 1 Overview For the practical exercises you have got a TUTORIAL, with which you can work most of the time on your own. There
More informationUsing an HPC Cloud for Weather Science
Using an HPC Cloud for Weather Science Provided By: Transforming Operational Environmental Predictions Around the Globe Moving EarthCast Technologies from Idea to Production EarthCast Technologies produces
More informationICON for HD(CP) 2. High Definition Clouds and Precipitation for Advancing Climate Prediction
ICON for HD(CP) 2 High Definition Clouds and Precipitation for Advancing Climate Prediction High Definition Clouds and Precipitation for Advancing Climate Prediction ICON 2 years ago Parameterize shallow
More informationMeteorology and Python
Meteorology and Python desperately trying to forget technical details Claude Gibert, Europython 2011 Background Meteorology - NWP Numerical Weather Prediction ECMWF European Centre for Medium-Range Weather
More informationHortonworks Data Platform
Hortonworks Data Platform Workflow Management (August 31, 2017) docs.hortonworks.com Hortonworks Data Platform: Workflow Management Copyright 2012-2017 Hortonworks, Inc. Some rights reserved. The Hortonworks
More information5.3 Install grib_api for OpenIFS
5.3 Install grib_api for OpenIFS Introduction The ECMWF grib_api software library provides a set of functions/subroutines and command line tools for encoding and decoding WMO FM- 92 GRIB edition 1 and
More informationHPC Input/Output. I/O and Darshan. Cristian Simarro User Support Section
HPC Input/Output I/O and Darshan Cristian Simarro Cristian.Simarro@ecmwf.int User Support Section Index Lustre summary HPC I/O Different I/O methods Darshan Introduction Goals Considerations How to use
More informationAutomating Real-time Seismic Analysis
Automating Real-time Seismic Analysis Through Streaming and High Throughput Workflows Rafael Ferreira da Silva, Ph.D. http://pegasus.isi.edu Do we need seismic analysis? Pegasus http://pegasus.isi.edu
More informationPython: Working with Multidimensional Scientific Data. Nawajish Noman Deng Ding
Python: Working with Multidimensional Scientific Data Nawajish Noman Deng Ding Outline Scientific Multidimensional Data Ingest and Data Management Analysis and Visualization Extending Analytical Capabilities
More informationIntroduction to ECMWF resources:
Introduction to ECMWF resources: Computing and archive services. and how to access them Paul Dando User Support Paul.Dando@ecmwf.int advisory@ecmwf.int University of Reading - 23 January 2014 ECMWF Slide
More informationAnne Fouilloux. Fig. 1 Use of observational data at ECMWF since CMA file structure.
ODB (Observational Database) and its usage at ECMWF Anne Fouilloux Abstract ODB stands for Observational DataBase and has been developed at ECMWF since mid-1998 by Sami Saarinen. The main goal of ODB is
More informationSTARTING THE DDT DEBUGGER ON MIO, AUN, & MC2. (Mouse over to the left to see thumbnails of all of the slides)
STARTING THE DDT DEBUGGER ON MIO, AUN, & MC2 (Mouse over to the left to see thumbnails of all of the slides) ALLINEA DDT Allinea DDT is a powerful, easy-to-use graphical debugger capable of debugging a
More informationMERCED CLUSTER BASICS Multi-Environment Research Computer for Exploration and Discovery A Centerpiece for Computational Science at UC Merced
MERCED CLUSTER BASICS Multi-Environment Research Computer for Exploration and Discovery A Centerpiece for Computational Science at UC Merced Sarvani Chadalapaka HPC Administrator University of California
More informationTECHNICAL GUIDELINES FOR APPLICANTS TO PRACE 13 th CALL (T ier-0)
TECHNICAL GUIDELINES FOR APPLICANTS TO PRACE 13 th CALL (T ier-0) Contributing sites and the corresponding computer systems for this call are: BSC, Spain IBM System x idataplex CINECA, Italy Lenovo System
More informationCSinParallel Workshop. OnRamp: An Interactive Learning Portal for Parallel Computing Environments
CSinParallel Workshop : An Interactive Learning for Parallel Computing Environments Samantha Foley ssfoley@cs.uwlax.edu http://cs.uwlax.edu/~ssfoley Josh Hursey jjhursey@cs.uwlax.edu http://cs.uwlax.edu/~jjhursey/
More informationTools and Methodology for Ensuring HPC Programs Correctness and Performance. Beau Paisley
Tools and Methodology for Ensuring HPC Programs Correctness and Performance Beau Paisley bpaisley@allinea.com About Allinea Over 15 years of business focused on parallel programming development tools Strong
More informationSCITAS. Scientific IT and Application support unit at EPFL. created in February today : 5 system engineers + 6 application experts SCITAS
SCITAS Scientific IT and Application support unit at EPFL created in February 2014 today : 5 system engineers + 6 application experts SCITAS SCITAS activities responsible for the central EPFL computing
More informationParallel Computing Ideas
Parallel Computing Ideas K. 1 1 Department of Mathematics 2018 Why When to go for speed Historically: Production code Code takes a long time to run Code runs many times Code is not end in itself 2010:
More informationUsing EasyBuild and Continuous Integration for Deploying Scientific Applications on Large Scale Production Systems
Using EasyBuild and Continuous Integration for Deploying Scientific Applications on Large HPC Advisory Council Swiss Conference Guilherme Peretti-Pezzi, CSCS April 11, 2017 Table of Contents 1. Introduction:
More informationAdvanced Parallel Programming
Sebastian von Alfthan Jussi Enkovaara Pekka Manninen Advanced Parallel Programming February 15-17, 2016 PRACE Advanced Training Center CSC IT Center for Science Ltd, Finland All material (C) 2011-2016
More informationParallel, In Situ Indexing for Data-intensive Computing. Introduction
FastQuery - LDAV /24/ Parallel, In Situ Indexing for Data-intensive Computing October 24, 2 Jinoh Kim, Hasan Abbasi, Luis Chacon, Ciprian Docan, Scott Klasky, Qing Liu, Norbert Podhorszki, Arie Shoshani,
More informationCDI-pio & XIOS I/O servers compatibility with HR climate models Eric Maisonnave, Irina Fast, Thomas Jahns, Joachim Biercamp, Stéphane Sénési, Yann
CDI-pio & XIOS I/O servers compatibility with HR climate models Eric Maisonnave, Irina Fast, Thomas Jahns, Joachim Biercamp, Stéphane Sénési, Yann Meurdesoif, Uwe Fladrich TR/CMGC/17/52 Abstract I/O performance
More informationNetCDF-4: : Software Implementing an Enhanced Data Model for the Geosciences
NetCDF-4: : Software Implementing an Enhanced Data Model for the Geosciences Russ Rew, Ed Hartnett, and John Caron UCAR Unidata Program, Boulder 2006-01-31 Acknowledgments This work was supported by the
More informationCLAW FORTRAN Compiler source-to-source translation for performance portability
CLAW FORTRAN Compiler source-to-source translation for performance portability XcalableMP Workshop, Akihabara, Tokyo, Japan October 31, 2017 Valentin Clement valentin.clement@env.ethz.ch Image: NASA Summary
More informationWELCOME to the PRACTICAL EXERCISES
WELCOME to the PRACTICAL EXERCISES Overview For the practical exercises you have got a TUTORIAL, with which you can work most of the time on your own. There are 6 lessons, in which you will learn about
More informationPerformance Analysis of Parallel Scientific Applications In Eclipse
Performance Analysis of Parallel Scientific Applications In Eclipse EclipseCon 2015 Wyatt Spear, University of Oregon wspear@cs.uoregon.edu Supercomputing Big systems solving big problems Performance gains
More informationAn Introduction to Software Architecture By David Garlan & Mary Shaw 94
IMPORTANT NOTICE TO STUDENTS These slides are NOT to be used as a replacement for student notes. These slides are sometimes vague and incomplete on purpose to spark a class discussion An Introduction to
More informationBelle II - Git migration
Belle II - Git migration Why git? Stash GIT service managed by DESY Powerful branching and merging capabilities Resolution of (JIRA) issues directly be map to branches and commits Feature freeze in pre-release
More informationOpen Data Standards for Administrative Data Processing
University of Pennsylvania ScholarlyCommons 2018 ADRF Network Research Conference Presentations ADRF Network Research Conference Presentations 11-2018 Open Data Standards for Administrative Data Processing
More informationHybrid Model Parallel Programs
Hybrid Model Parallel Programs Charlie Peck Intermediate Parallel Programming and Cluster Computing Workshop University of Oklahoma/OSCER, August, 2010 1 Well, How Did We Get Here? Almost all of the clusters
More informationThe Future of Interconnect Technology
The Future of Interconnect Technology Michael Kagan, CTO HPC Advisory Council Stanford, 2014 Exponential Data Growth Best Interconnect Required 44X 0.8 Zetabyte 2009 35 Zetabyte 2020 2014 Mellanox Technologies
More informationDynamic Cuda with F# HPC GPU & F# Meetup. March 19. San Jose, California
Dynamic Cuda with F# HPC GPU & F# Meetup March 19 San Jose, California Dr. Daniel Egloff daniel.egloff@quantalea.net +41 44 520 01 17 +41 79 430 03 61 About Us! Software development and consulting company!
More informationThe Evolution of Big Data Platforms and Data Science
IBM Analytics The Evolution of Big Data Platforms and Data Science ECC Conference 2016 Brandon MacKenzie June 13, 2016 2016 IBM Corporation Hello, I m Brandon MacKenzie. I work at IBM. Data Science - Offering
More informationAddressing the Increasing Challenges of Debugging on Accelerated HPC Systems. Ed Hinkel Senior Sales Engineer
Addressing the Increasing Challenges of Debugging on Accelerated HPC Systems Ed Hinkel Senior Sales Engineer Agenda Overview - Rogue Wave & TotalView GPU Debugging with TotalView Nvdia CUDA Intel Phi 2
More informationICON Performance Benchmark and Profiling. March 2012
ICON Performance Benchmark and Profiling March 2012 Note The following research was performed under the HPC Advisory Council activities Participating vendors: Intel, Dell, Mellanox Compute resource - HPC
More informationREQUEST FOR A SPECIAL PROJECT
REQUEST FOR A SPECIAL PROJECT 2018 2020 MEMBER STATE: Germany, Greece, Italy This form needs to be submitted via the relevant National Meteorological Service. Principal Investigator 1 Amalia Iriza (NMA,Romania)
More informationAdvances in Time-Parallel Four Dimensional Data Assimilation in a Modular Software Framework
Advances in Time-Parallel Four Dimensional Data Assimilation in a Modular Software Framework Brian Etherton, with Christopher W. Harrop, Lidia Trailovic, and Mark W. Govett NOAA/ESRL/GSD 28 October 2016
More informationAMSC/CMSC 662 Computer Organization and Programming for Scientific Computing Fall 2011 Introduction to Parallel Programming Dianne P.
AMSC/CMSC 662 Computer Organization and Programming for Scientific Computing Fall 2011 Introduction to Parallel Programming Dianne P. O Leary c 2011 1 Introduction to Parallel Programming 2 Why parallel
More informationMetview s new Python interface
Metview s new Python interface Workshop on developing Python frameworks for earth system sciences. ECMWF, 2018 Iain Russell Development Section, ECMWF Thanks to Sándor Kertész Fernando Ii Stephan Siemen
More informationUsing ODB at ECMWF. Piotr Kuchta Sándor Kertész. Development Section ECMWF. Slide 1. MOS Workshop, 2013 November 18-20, ECMWF
Using ODB at ECMWF Piotr Kuchta Sándor Kertész Development Section ECMWF Slide 1 MOS Workshop, 2013 November 18-20, ECMWF 1 History of ODB in a nutshell 1998 2008, Sami Saarinen Database of observations
More informationIntroduction to High-Performance Computing
Introduction to High-Performance Computing Dr. Axel Kohlmeyer Associate Dean for Scientific Computing, CST Associate Director, Institute for Computational Science Assistant Vice President for High-Performance
More informationIntroduction to parallel Computing
Introduction to parallel Computing VI-SEEM Training Paschalis Paschalis Korosoglou Korosoglou (pkoro@.gr) (pkoro@.gr) Outline Serial vs Parallel programming Hardware trends Why HPC matters HPC Concepts
More informationHerschel Data Analysis Guerilla Style: Keeping flexibility in a system with long development cycles. Bernhard Schulz
GRITS 17 June 2011 Herschel Data Analysis Guerilla Style: Keeping flexibility in a system with long development cycles Bernhard Schulz NASA Herschel Science Center HCSS The Herschel Common Science System
More informationMetview 4 ECMWF s next generation meteorological workstation
Metview 4 ECMWF s next generation meteorological workstation Iain Russell Graphics Section ECMWF Slide 1 21 st EGOWS, Reading, 1 4 June 2010 1 What is Metview? (1) Working environment for Operational and
More informationThe CEDA Web Processing Service for rapid deployment of earth system data services
The CEDA Web Processing Service for rapid deployment of earth system data services Stephen Pascoe Ag Stephens Phil Kershaw Centre of Environmental Data Archival 1 1 Overview of CEDA-WPS History first implementation
More informationDiskBoss DATA MANAGEMENT
DiskBoss DATA MANAGEMENT File Synchronization Version 9.1 Apr 2018 www.diskboss.com info@flexense.com 1 1 DiskBoss Overview DiskBoss is an automated, policy-based data management solution allowing one
More informationGeneral Purpose GPU Programming. Advanced Operating Systems Tutorial 9
General Purpose GPU Programming Advanced Operating Systems Tutorial 9 Tutorial Outline Review of lectured material Key points Discussion OpenCL Future directions 2 Review of Lectured Material Heterogeneous
More informationECMWF s Next Generation IO for the IFS Model
ECMWF s Next Generation IO for the Model Part of ECMWF s Scalability Programme Tiago Quintino, B. Raoult, P. Bauer ECMWF tiago.quintino@ecmwf.int ECMWF January 14, 2016 ECMWF s HPC Targets What do we do?
More informationCHEP 2013 October Amsterdam K De D Golubkov A Klimentov M Potekhin A Vaniachine
Task Management in the New ATLAS Production System CHEP 2013 October 14-18 K De D Golubkov A Klimentov M Potekhin A Vaniachine on behalf of the ATLAS Collaboration Overview The ATLAS Production System
More informationMARS Code Reorganization
MARS Code Reorganization Build Systems, Code Repositories, and much more... Tiago Quintino Sebastien Villaume, Manuel Fuentes, Baudouin Raoult mars-admins@ecmwf.int ECMWF March 9, 2016 Code Cleanup MARS
More informationComputer Architectures and Aspects of NWP models
Computer Architectures and Aspects of NWP models Deborah Salmond ECMWF Shinfield Park, Reading, UK Abstract: This paper describes the supercomputers used in operational NWP together with the computer aspects
More informationGNU Radio Technical Update
GNU Radio Technical Update Johnathan Corgan GRCON17 GRCON17 GNU GNU Radio Radio Technical Technical Update Update September September 2017 2017 Topics Release 3.8 Features and capabilities Milestones and
More informationTigres: Template Interfaces for Agile Parallel Data-Intensive Science
Tigres: Template Interfaces for Agile Parallel Data-Intensive Science http://tigres.lbl.gov Lavanya Ramakrishnan LRamakrishnan@lbl.gov 1 (CS Biased) View of Workflow Challenges: Gene2Life Molecular Biology
More informationA Parallelized Cartographic Modeling Language (CML)-Based Solution for Flux Footprint Modeling
A Parallelized Cartographic Modeling Language (CML)-Based Solution for Flux Footprint Modeling Michael E. Hodgson, April Hiscox, Shaowen Wang, Babak Behzad, Sara Flecher, Kiumars Soltani, Yan Liu and Anand
More informationIntroduction to Parallel Programming
Introduction to Parallel Programming David Lifka lifka@cac.cornell.edu May 23, 2011 5/23/2011 www.cac.cornell.edu 1 y What is Parallel Programming? Using more than one processor or computer to complete
More informationAllinea Unified Environment
Allinea Unified Environment Allinea s unified tools for debugging and profiling HPC Codes Beau Paisley Allinea Software bpaisley@allinea.com 720.583.0380 Today s Challenge Q: What is the impact of current
More informationTowards Zero Cost I/O:
Towards Zero Cost I/O: Met Office Unified Model I/O Server Martyn Foster Martyn.Foster@metoffice.gov.uk A reminder Amdahl s Law Performance is always constrained by the serial code T T s T P T T s T 1
More informationIntro to Parallel Computing
Outline Intro to Parallel Computing Remi Lehe Lawrence Berkeley National Laboratory Modern parallel architectures Parallelization between nodes: MPI Parallelization within one node: OpenMP Why use parallel
More informationORAP Forum October 10, 2013
Towards Petaflop simulations of core collapse supernovae ORAP Forum October 10, 2013 Andreas Marek 1 together with Markus Rampp 1, Florian Hanke 2, and Thomas Janka 2 1 Rechenzentrum der Max-Planck-Gesellschaft
More informationPython ecosystem for scientific computing with ABINIT: challenges and opportunities. M. Giantomassi and the AbiPy group
Python ecosystem for scientific computing with ABINIT: challenges and opportunities M. Giantomassi and the AbiPy group Frejus, May 9, 2017 Python package for: generating input files automatically post-processing
More informationThe Cylc Suite Engine User Guide
The Cylc Suite Engine User Guide 7.6.1 Released Under the GNU GPL v3.0 Software License Copyright (C) 2008-2018 NIWA Hilary Oliver March 28, 2018 Abstract Cylc ( silk ) is a metascheduler 1 for cycling
More informationNext-Generation Cloud Platform
Next-Generation Cloud Platform Jangwoo Kim Jun 24, 2013 E-mail: jangwoo@postech.ac.kr High Performance Computing Lab Department of Computer Science & Engineering Pohang University of Science and Technology
More informationDatabase infrastructure for electronic structure calculations
Database infrastructure for electronic structure calculations Fawzi Mohamed fawzi.mohamed@fhi-berlin.mpg.de 22.7.2015 Why should you be interested in databases? Can you find a calculation that you did
More informationExperiences from the Fortran Modernisation Workshop
Experiences from the Fortran Modernisation Workshop Wadud Miah (Numerical Algorithms Group) Filippo Spiga and Kyle Fernandes (Cambridge University) Fatima Chami (Durham University) http://www.nag.co.uk/content/fortran-modernization-workshop
More informationMetview 4 ECMWF s latest generation meteorological workstation
Metview 4 ECMWF s latest generation meteorological workstation Iain Russell, Stephan Siemen, Fernando Ii, Sándor Kertész, Sylvie Lamy-Thépaut, Vesa Karhila Version 4 builds on the flexible and proven modular
More informationNational Aeronautics and Space and Administration Space Administration. CFE CMake Build System
National Aeronautics and Space and Administration Space Administration CFE CMake Build System 1 1 Simplify integrating apps together CFS official Recycled from other projects Custom LC... SC HK A C B Z
More informationThe challenges of new, efficient computer architectures, and how they can be met with a scalable software development strategy.! Thomas C.
The challenges of new, efficient computer architectures, and how they can be met with a scalable software development strategy! Thomas C. Schulthess ENES HPC Workshop, Hamburg, March 17, 2014 T. Schulthess!1
More informationAMath 483/583 Lecture 2. Notes: Notes: Homework #1. Class Virtual Machine. Notes: Outline:
AMath 483/583 Lecture 2 Outline: Binary storage, floating point numbers Version control main ideas Client-server version control, e.g., CVS, Subversion Distributed version control, e.g., git, Mercurial
More informationNew programming languages and paradigms
New programming languages and paradigms Seminar: Neueste Trends im Hochleistungsrechnen Arbeitsbereich Wissenschaftliches Rechnen Fachbereich Informatik Fakultät für Mathematik, Informatik und Naturwissenschaften
More informationAccelerate HPC Development with Allinea Performance Tools
Accelerate HPC Development with Allinea Performance Tools 19 April 2016 VI-HPS, LRZ Florent Lebeau / Ryan Hulguin flebeau@allinea.com / rhulguin@allinea.com Agenda 09:00 09:15 Introduction 09:15 09:45
More informationECE 3574: Applied Software Design
ECE 3574: Applied Software Design Chris Wyatt Spring 2018 Welcome to ECE 3574: Applied Software Design CRN 19016 Website: https://filebox.ece.vt.edu/~ece3574 Instructor: Chris Wyatt, clwyatt@vt.edu Today
More information