Skip to main content

Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing

Deliverables

Concepts for READEX Tool Suite

A joint report describing the concepts for the integrated READEX Tool Suite. It includes the following parts that will be contributed by the leaders of the respective WPs: (1) A description of available tuning parameters (WP1), (2) The overall approach for scenario identification and the pre-computation of optimal platform configurations (WP2), (3) The architec-ture of the READEX Runtime Library (RRL) and the run-time scenario detection (WP3) (4) The concepts for the tool-suite integration (WP4)

Updated report on dissemination, communication, standardisation, and exploitation activities

This is an updated report of D6.3. This report will further document the final plans for short-term, mid-term and long-term exploitation and dis-semination of project results.

Future Parameters

A report describing system parameters that are available at the end of the project life-time in addition to the parameters described in D1.1. The re-port also contains an outlook on potential future parameters that would help the auto-tuning efforts.

Final description of the READEX Programming Paradigm

A report describing the final version of the new READEX Programming Paradigm that allow users to provide application-domain knowledge to the auto-tuning process, both to mark alternative code-paths and to support the detection and identification of scenarios at design-time and at run-time.

Second Periodic Report

This deliverable reports on the achieved results of the project and the overall schedule of tasks. It reports any deviations from the work plan and risks occurred with the performed work. It provides the final status of the REDEAX financials.

Report on application dynamism

A report describing and quantifying the different types of dynamic be-haviour found in the selected benchmarks and applications.

Initial report on dissemination, communication, standardisation, and exploitation activities

This initial report describes the participation in events for dissemination, communication, and exploitation. The report also provides a review of the main initiatives conducted by working groups and forums recognised as of interest from standardisation point of view. Possible participation in standardisation initiatives and related results, if any, will also be re-ported in this deliverable, together with any other results achieved by dissemination, communication, and exploitation activities.

Visualization of dynamism

Deliver the integration of the information on application dynamism dis-covered by PTF in Score-P to facilitate the visualization of dynamic behaviour / application phases in the Vampir trace visualization tool.

Report on external advisory board creation

A report describing the members of the External Advisory Board (EAB).

Analysis of tuning potential and scenario identification

The report will present the analysis strategy to predict the suitability of the application for the READEX tuning approach as well as the approach taken for scenario identification.

Final mechanisms for run-time detection, switching, and calibration

The final automatic mechanisms for inclusion in the RRL.

First Periodic Report

This deliverable reports on the achieved results of the project and the overall schedule of tasks. It reports any deviations from the work plan and risks associated with the performed work and planned milestones. It provides further a status of the READEX financial status and its budget spending.

Final RRL architecture

A report describing the principles for the scalable run-time library.

Dissemination plan

The dissemination plan has been established and is in operation

Final computation of configurations

Covers the final implementation of the design time computation of optimized system configurations.

Final tuning plugins

A set of tuning plugins for hardware, system-software and application parameters. Final version.

Evaluation of the READEX Tool Suite using the READEX test-suite

Final evaluation of the READEX Tool Suite using the READEX test-suite delivered in D5.2.

Final Report

This final report summarizes the project’s activities over its full duration. This report will cover the main aspects of the work, objectives, results and conclusions.

READEX Project Management Tools

An internal website will be created initially to form a repository of documentary and other information that is relevant to the implementation of the project, and that can also be used to provide an initial intra-project communication system. Additionally, the usual collaboration tools and procedures (e.g. mailing list, wiki, SVN) will be installed and made available to the project partners.

Enhanced READEX Public Website

The enhanced version of the project website will additionally contain the specific section targeted to the end users.

Initial READEX Public Website

The initial version of the website will contain the collaboration and communication tools and it will describe the vision of the project as well as the structure and role of the partners.

Final READEX Tool Suite

The final version of the integrated READEX Tool Suite to be publicly re-leased.

Prototype READEX Tool Suite

A prototype of the integrated READEX Tool Suite for first testing and validation

Hardware and system-software tuning plugins

A set of tuning plugins for hardware and system-software. Preliminary results for application parameters.

READEX Tool Suite version 2

A second version of the integrated READEX Tool Suite to be used for testing and validation.

Extended READEX test-suite with manually tuned applications

The READEX test-suite extended with manually tuned benchmarks and application to be used for comparison of productivity (required effort and achieved improvement in energy-efficiency) with the READEX Tool Suite.

Publications

Towards fine-grained dynamic tuning of HPC applications on modern multi-core architectures

Author(s): Mohammed Sourouri, Espen Birger Raknes, Nico Reissmann, Johannes Langguth, Daniel Hackenberg, Robert Schöne, Per Gunnar Kjeldsberg
Published in: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis on - SC '17, 2017, Page(s) 1-12
DOI: 10.1145/3126908.3126945

A Simple Framework for Energy Efficiency Evaluation and Hardware Parameter Tuning with Modular Support for Different HPC Platforms

Author(s): Ondrej Vysocky, Jan Zapletal and Lubomir Riha
Published in: 2018

Extending the Functionality of Score-P through Plugins: Interfaces and Use Cases

Author(s): Robert Schöne, Ronny Tschüter, Thomas Ilsche, Joseph Schuchart, Daniel Hackenberg, Wolfgang E. Nagel
Published in: 2017, Page(s) 59-82
DOI: 10.1007/978-3-319-56702-0_4

Optimal Hardware Parameters Prediction for Best Energy-to-Solution of Sparse Matrix Operations Using Machine Learning Techniques

Author(s): Vojtech Nikl, Ondrej Vysocky, Lubomir Riha, Jan Zapletal
Published in: 2018

READEX: Linking two ends of the computing continuum to improve energy-efficiency in dynamic applications

Author(s): Per Gunnar Kjeldsberg, Andreas Gocht, Michael Gerndt, Lubomir Riha, Joseph Schuchart, Umbreen Sabir Mian
Published in: Design, Automation & Test in Europe Conference & Exhibition (DATE), 2017, 2017, Page(s) 109-114
DOI: 10.23919/DATE.2017.7926967

READEX Tool Suite for Energy-efficiency Tuning of HPC Applications

Author(s): Anamika Chowdhury, Madhura Kumaraswamy, Michael Gerndt
Published in: Proceedings of the 2017 Workshop on Software Engineering Methods for Parallel and High Performance Applications - SEM4HPC '17, 2017, Page(s) 11-12
DOI: 10.1145/3085158.3091994

Detecting Memory-Boundedness with Hardware Performance Counters

Author(s): Daniel Molka, Robert Schöne, Daniel Hackenberg, Wolfgang E. Nagel
Published in: Proceedings of the 8th ACM/SPEC on International Conference on Performance Engineering - ICPE '17, 2017, Page(s) 27-38
DOI: 10.1145/3030207.3030223

System Monitoring with lo2s: Power and Runtime Impact of C-State Transitions

Author(s): Thomas Ilsche, Robert Schone, Philipp Joram, Mario Bielert, Andreas Gocht
Published in: 2018 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), 2018, Page(s) 712-715
DOI: 10.1109/IPDPSW.2018.00114

Domain Knowledge Specification for Energy Tuning

Author(s): Chowdhury, Anamika; Kumaraswamy, Madhura; Gerndt, Michael; Bendifallah, Zakaria; Bouizi, Othman; Řı́ha, Lubomı́r; Vysocký, Ondřej; Beseda, Martin; Zapletal, Jan
Published in: Issue 4, 2017
DOI: 10.5281/zenodo.815851

lo2s — Multi-core System and Application Performance Analysis for Linux

Author(s): Thomas Ilsche, Robert Schone, Mario Bielert, Andreas Gocht, Daniel Hackenberg
Published in: 2017 IEEE International Conference on Cluster Computing (CLUSTER), 2017, Page(s) 801-804
DOI: 10.1109/CLUSTER.2017.116

Powernightmares: The challenge of efficiently using sleep states on multi-core systems

Author(s): Thomas Ilsche, Marcus Hähnel, Robert Schöne, Mario Bielert, Daniel Hackenberg
Published in: 2018, Page(s) 623-635
DOI: 10.1007/978-3-319-75178-8_50

Energy consumption optimization of the total-FETI solver by changing the CPU frequency

Author(s): David Horak, Lubomir Riha, Radim Sojka, Jakub Kruzik, Martin Beseda, Martin Cermak, Joseph Schuchart
Published in: AIP Conference Proceedings volume 1863, 2017, Page(s) 340004
DOI: 10.1063/1.4992511

MERIC and RADAR Generator: Tools for Energy Evaluation and Runtime Tuning of HPC Applications

Author(s): Ondrej Vysocky, Martin Beseda, Lubomír Říha, Jan Zapletal, Michael Lysaght, Venkatesh Kannan
Published in: 2018, Page(s) 144-159
DOI: 10.1007/978-3-319-97136-0_11

Evaluation of the HPC Applications Dynamic Behavior in Terms of Energy Consumption

Author(s): O. Vysocky, M. Beseda, L. Riha, J. Zapletal, V. Nikl, M. Lysaght, V. Kannan
Published in: Proceedings of the Fifth International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering, 2017
DOI: 10.4203/ccp.111.3

Leveraging Inter-Phase Application Dynamism for Energy-Efficiency Auto-tuning

Author(s): M. Kumaraswamy, M. Gerndt
Published in: Issue 1, 2018

Diagnosing Highly-Parallel OpenMP Programs with Aggregated Grain Graphs

Author(s): Nico Reissmann, Ananya Muddukrishna
Published in: 2018, Page(s) 106-119
DOI: 10.1007/978-3-319-96983-1_8

Software Controlled Clock Modulation for Energy Efficiency Optimization on Intel Processors

Author(s): Robert Schone, Thomas Ilsche, Mario Bielert, Daniel Molka, Daniel Hackenberg
Published in: 2016 4th International Workshop on Energy Efficient Supercomputing (E2SC), 2016, Page(s) 69-76
DOI: 10.1109/E2SC.2016.015

Solving Contact Mechanics Problems with PERMON

Author(s): Vaclav Hapla, David Horak, Lukas Pospisil, Martin Cermak, Alena Vasatova, Radim Sojka
Published in: 2015, Page(s) 101-115
DOI: 10.1007/978-3-319-40361-8_7

The READEX Project for Dynamic Energy Efficiency Tuning

Author(s): Michael Gerndt
Published in: Proceedings of the ACM Workshop on Software Engineering Methods for Parallel and High Performance Applications - SEM4HPC '16, 2016, Page(s) 11-12
DOI: 10.1145/2916026.2916033

On the performance and energy efficiency of the PGAS programming model on multicore architectures

Author(s): Jeremie Lagraviere, Johannes Langguth, Mohammed Sourouri, Phuong H. Ha, Xing Cai
Published in: 2016 International Conference on High Performance Computing & Simulation (HPCS), 2016, Page(s) 800-807
DOI: 10.1109/HPCSim.2016.7568416

Autotuning of MPI Applications Using PTF

Author(s): Anna Sikora, Eduardo César, Isaías Comprés, Michael Gerndt
Published in: Proceedings of the ACM Workshop on Software Engineering Methods for Parallel and High Performance Applications - SEM4HPC '16, 2016, Page(s) 31-38
DOI: 10.1145/2916026.2916028

Accelerating 3D Elastic Wave Equations on Knights Landing based Intel Xeon Phi processors

Author(s): M. Sourouri, E. B. Raknes
Published in: 2017

Run-Time Exploitation of Application Dynamism for Energy-Efficient Exascale Computing (READEX)

Author(s): Yury Oleynik, Michael Gerndt, Joseph Schuchart, Per Gunnar Kjeldsberg, Wolfgang E. Nagel
Published in: 2015 IEEE 18th International Conference on Computational Science and Engineering, 2015, Page(s) 347-350
DOI: 10.1109/CSE.2015.55

The energy consumption optimization of the BLAS routines

Author(s): Radim Sojka, Lubomir Riha, David Horak, Jakub Kruzik, Martin Beseda, Martin Cermak
Published in: AIP Conference Proceedings volume 1863, 2017, Page(s) 340015
DOI: 10.1063/1.4992522

Energy consumption optimization of the Total-FETI solver and BLAS routines by changing the CPU frequency

Author(s): David Horak, Lubomir Riha, Radim Sojka, Jakub Kruzik, Martin Beseda
Published in: 2016 International Conference on High Performance Computing & Simulation (HPCS), 2016, Page(s) 1031-1032
DOI: 10.1109/HPCSim.2016.7568453

Investigating and Exploiting Application Dynamism For Energy-Efficient Exascale Computing

Author(s): Venkatesh Kannana, Lubomı́r Ř́ıha, Michael Gerndt, Anamika Chowdhury, Ondrej Vysocky, Martin Beseda, Horák David, Radim Sojka, Jakub Kruzík, Michael Lysaght
Published in: 2017
DOI: 10.5281/zenodo.832027

READEX gjør dynamiske HPC-applikasjoner energieffektive

Author(s): Per Gunnar Kjeldsberg
Published in: 2018

A massively parallel and memory-efficient FEM toolbox with a hybrid total FETI solver with accelerator support

Author(s): Lubomir Riha, Michal Merta, Radim Vavrik, Tomas Brzobohaty, Alexandros Markopoulos, Ondrej Meca, Ondrej Vysocky, Tomas Kozubek, Vit Vondrak
Published in: The International Journal of High Performance Computing Applications, 2018, Page(s) 109434201879845, ISSN 1094-3420
DOI: 10.1177/1094342018798452

The shift from processor power consumption to performance variations: fundamental implications at scale

Author(s): Joseph Schuchart, Daniel Hackenberg, Robert Schöne, Thomas Ilsche, Ramkumar Nagappan, Michael K. Patterson
Published in: Computer Science - Research and Development, Issue 31/4, 2016, Page(s) 197-205, ISSN 1865-2034
DOI: 10.1007/s00450-016-0327-2

Design-time Analysis for the READEX Tool Suite

Author(s): M. Kumaraswamy, A. Chowdhury, M. Gerndt
Published in: IOS Press, Issue 2, 2018, ISSN 0927-5452
DOI: 10.3233/978-1-61499-843-3-307

Domain knowledge specification for energy tuning

Author(s): Madhura Kumaraswamy, Anamika Chowdhury, Michael Gerndt, Zakaria Bendifallah, Othman Bouizi, Uldis Locans, Lubomír Říha, Ondřej Vysocký, Martin Beseda, Jan Zapletal
Published in: Concurrency and Computation: Practice and Experience, 2018, Page(s) e4650, ISSN 1532-0626
DOI: 10.1002/cpe.4650

The READEX formalism for automatic tuning for energy efficiency

Author(s): Joseph Schuchart, Michael Gerndt, Per Gunnar Kjeldsberg, Michael Lysaght, David Horák, Lubomír Říha, Andreas Gocht, Mohammed Sourouri, Madhura Kumaraswamy, Anamika Chowdhury, Magnus Jahre, Kai Diethelm, Othman Bouizi, Umbreen Sabir Mian, Jakub Kružík, Radim Sojka, Martin Beseda, Venkatesh Kannan, Zakaria Bendifallah, Daniel Hackenberg, Wolfgang E. Nagel
Published in: Computing, 2017, ISSN 0010-485X
DOI: 10.1007/s00607-016-0532-7

Tools for assessing and optimizing the energy requirements of high performance scientific computing software

Author(s): Kai Diethelm
Published in: PAMM, Issue 16/1, 2016, Page(s) 837-838, ISSN 1617-7061
DOI: 10.1002/pamm.201610407

Runtime Exploitation of Application Dynamism for Energy-efficient Exascale Computing

Author(s): P.G. Kjeldsberg, R. Schöne, M. Gerndt, L. Riha, V. Kannan, K. Diethelm, M-C. Sawley, J.Zapletal, O. Vysocky, M.Kumaraswamy, and W.E. Nagel
Published in: System Scenario-based Design Principles and Applications, 2018

Software

Readex-Apps Repository

Author(s): Zapletal, Jan; Vysocky, Ondrej; Riha, Lubomir; Kannan, Venkatesh; Madhura Kumaraswamy; Chowdhury, Anamika; Gerndt, Michael; Diethelm, Kai; Schöne, Robert
DOI: 10.5281/zenodo.1451437; 10.5281/zenodo.1451438
Publisher: Zenodo