DEEP – SOFTWARE FOR EXASCALE ARCHITECTURES | DEEP-SEA | Project | Results | H2020 | CORDIS

Risultati finali

Definition of the quality control processes and templates for internal verification and document review for all project results and deliverables

Interoperability development

Detailed report on proven ability of GPISpace to support the execution of MPI or GASPI programs spanning over multiple nodes and detailed performance study of MPIGPU performance bottleneck

Application use cases and traces

Use cases integrated in JUBE profile and trace files provided to other WPs

Final evaluation of the system software stack

Optimisations, adjustments and bug fixing in accordance with the user experiences.This Deliverable is led by ParTec.

Initial application co-design input

Documents requirements of all applications for codesign includes all data analytics SW requirements compute memory communication performance footprints communication patterns etc

Final report on applications experience

Reports the results of the application experience using the DEEP-SEA developments. Performance and efficiency (both according to the most relevant metric for the application), scalability, and portability will be measured and compared with the code performances at the beginning of the project.

Applications use of DEEP-SEA software stack

Details the tools SW components and programming models that each application will use with further codesign input on the needed functionalities

Software specification

Specification of the complete SW stack based on the requirements collected from WPs 1 to 5 This comprises WP3internal interfaces and dependencies as well as the interplay with WPs 2 4 and 5

Documentation of last improvements of the tools

Documentation including bug fixes, lessons learned from deployment and next steps.

MSA-driven extension to system-wide programming and new memory integration

Detailed report and intermediate release on extension of MPI libraries, including collective communications, RMA communication, tuning, and the MPI memory management extension. Also includes details on NVM as a fast buffer for a workflow’s intermediate data, and GPI-2 extension to support persistent segments.

Report on standardisation activities

Summarising report covering all proposals made to the mentioned standardisation bodies based on work in DEEP-SEA as well as open standardisation potentials to create future roadmaps.

Final report

Description of the technical and scientific results of the project

Repository for training material

Design, access information and initial contents for the training repository.

Intermediate node-level programming environment

Intermediate release and detailed report of programming environment. Final release and SW stack for processing in memory (PIM). This SW release will provide the programming environment to the application partners for their work leading to D1.4 First evaluation results.

Complete system software implementation

Full system SW implementation of all components of the system SW stack meeting the requirements collected in D3.1 and matching the final specification developed in D3.2.This Deliverable is led by ParTec.

Final Node-level programming environment

Final release and detailed report of programming environment and basis for D1.5 Final report on applications experience. Any remaining integration and debugging work will continue in the integration task

Initial node-level programming environment

Initial release and detailed report of node-level programming environment. This SW release will enable early work by the application partners

Resiliency support

Intermediate release and detailed report on resiliency enhancement. It includes description on resiliency in MPI sessions, slum extensions, Support of persistent segments in the GPI-2 implementation.

Malleability concept and early prototype

Intermediate release and detailed report on malleability including concept and prototype version of MPIsessionbased malleability within MPI

Final system-level programming environment

Final release and detailed report on WP5 contributions to system-level programming environment. This report includes the malleability support (including MPI and OmpSs-2@cluster extensions), interoperability (including accelerator optimised communication, MPC+GPI interaction, ParaStation MPI communication layer support in GPI-2, GPI-Space programs that support MPI and GASPI at the same time, and both MPI+OmpSs-2@cluster and GPI+OmpSs-2@cluster), and resiliency capabilities (including support for persistent segments in in-memory checkpointing library, evaluation of Slurm extension, evaluation of the extension of the checkpoint restart capabilities, evaluation of Open MPI runtime fault tolerance). These evaluations will be conducted using micro-benchmark and applications from WP1.

Final release of the tools

Release with final feature set, tested and including updates to user documentation and installation guide.

Resource management and tool interface

Intermediate release and detailed report on the interfaces developed between MPI libraries and their collocated environment such as the resource manager and external tools.

Final outreach activity

Final brochure or project video, or an event or a combination of these. It will be decided during the project, which will be deemed most appropriate.

Communication and data management plan, toolkit and owned channels

Details of the communication and brand strategy including a toolkit with materials like logos and collateral templates and DEEPSEA communication channels It contains also the data management plan as part of the Open Research Data Pilot ORDP

Pubblicazioni

A Compiler Approach to Automatic Multi-Pumping

Autori: Johnsen, Carl-Johannes; De Matteis, Tiziano; Ben-Nun, Tal; Licht, Johannes de Fine; Hoefler, Torsten
Pubblicato in: Proceedings of the 41st IEEE/ACM International Conference on Computer-Aided Design, Numero 2, 2022
Editore: ACM
DOI: 10.1145/3508352.3549374

An Emulation Layer for Dynamic Resources with MPI Sessions

Autori: Jan Fecht; Martin Schreiber; Martin Schulz; Howard Pritchard; Daniel J. Holmes
Pubblicato in: Lecture Notes in Computer Science, Numero 2, 2023, ISBN 9783031232190
Editore: Springer
DOI: 10.1007/978-3-031-23220-6_10

Artifact and instructions to generate experimental results for Euro-Par 2022 paper: OmpSs-2@Cluster: Distributed memory execution of nested OpenMP-style tasks

Autori: Aguilar Mena, Jimmy; Shaaban, Omar; Beltran, Vicenç; Carpenter, Paul; Ayguadé, Eduard; Labarta Mancho, Jesus
Pubblicato in: Lecture Notes in Computer Science, Numero 2, 2022
Editore: Springer
DOI: 10.6084/m9.figshare.19960721.v1

NPBench: A Benchmarking Suite for High-Performance NumPy

Autori: Alexandros Nikolaos Ziogas, Tal Ben-Nun, Timo Schneider, and Torsten Hoefler
Pubblicato in: 2021
Editore: ICS'21

OmpSs-2@Cluster: Distributed memory execution of nested OpenMP-style tasks.

Autori: J. Aguilar Mena, O. Shaaban, V. Beltran, P. Carpenter, E. Ayguade, and J. Labarta
Pubblicato in: Proceedings of Euro-Par 2022, 2022
Editore: Springer
DOI: 10.1007/978-3-031-12597-3_20

Towards Dynamic Resource Management with MPI Sessions and PMIx

Autori: Huber, Dominik; Streubel, Maximilian; Comprés, Isaías; Schulz, Martin; Schreiber, Martin; Pritchard, Howard
Pubblicato in: uroMPI/USA '22: Proceedings of the 29th European MPI Users' Group Meeting, Numero 2, 2022
Editore: ACM
DOI: 10.1145/3555819.3555856

A Data-Centric Optimization Framework for Machine Learning

Autori: Oliver Rausch; Tal Ben-Nun; Nikoli Dryden; Andrei Ivanov; Shigang Li; Torsten Hoefler
Pubblicato in: ICS '22: Proceedings of the 36th ACM International Conference on Supercomputing, 2022
Editore: Association for Computing Machinery
DOI: 10.48550/arxiv.2110.10802

Productive Performance Engineering for Weather and Climate Modeling with Python

Autori: T. Ben-Nun, L. Groner, F. Deconinck, T. Wicky, E. Davis, J. Dahm, O. Elbert, R. George, J. McGibbon, L. Trümper, E. Wu, O. Fuhrer, T. Schulthess, T. Hoefler
Pubblicato in: SC'22, 2022
Editore: -

Exploring the impact of node failures on the resource allocation for parallel jobs

Autori: Ioannis Vardas, Manolis Ploumidis, Manolis Marazakis
Pubblicato in: Proceedings of the 14th Resilience Workshop, held in conjunction with Euro-Par, 2021
Editore: JuSER

Building Blocks for Network-Accelerated Distributed File Systems

Autori: Salvatore Di Girolamo, Daniele De Sensi, Konstantin Taranov, Milos Malesevic, Maciej Besta, Timo Schneider, Severin Kistler, Torsten Hoefler
Pubblicato in: SC'22, 2022
Editore: SC'22
DOI: 10.48550/arxiv.2206.10007

Impact of Cache Coherence on the Performance of Shared-Memory based MPI Primitives: A Case Study for Broadcast on Intel Xeon Scalable Processors - Computational Artifacts

Autori: Katevenis, George; Ploumidis, Manolis; Marazakis, Manolis
Pubblicato in: 52nd International Conference on Parallel Processing (ICPP), Numero 2, 2023
Editore: ACM
DOI: 10.5281/zenodo.8074488

Accelerating Brain Simulations with the Fast Multipole Method

Autori: H. Nöttgen, F. Czappa, and F. Wolf
Pubblicato in: Proceedings of Euro-Par 2022, 2022
Editore: Springer
DOI: 10.1007/978-3-031-12597-3_24

A High-Fidelity Flow Solver for Unstructured Meshes on Field-Programmable Gate Arrays: Design, Evaluation, and Future Challenges.

Autori: Martin Karp, Artur Podobas, Tobias Kenter, Niclas Jansson, Christian Plessl, Philipp Schlatter, and Stefano Markidis
Pubblicato in: HPCAsia2022, 2022
Editore: ACM
DOI: 10.1145/3492805.3492808

Maximum Flows in Parametric Graph Templates

Autori: Tal Ben-Nun; Lukas Gianinazzi; Torsten Hoefler; Yishai Oltchik
Pubblicato in: Lecture Notes in Computer Science, Numero vol 13898, 2023, Pagina/e 97-111
Editore: Springer
DOI: 10.1007/978-3-031-30448-4_8

Boosting Performance Optimization with Interactive Data Movement Visualization

Autori: Philipp Schaad, Tal Ben-Nun, Torsten Hoefler
Pubblicato in: 2022
Editore: SC'22
DOI: 10.48550/arxiv.2207.07433

Breaking Down the Parallel Performance of GROMACS, a High-Performance Molecular Dynamics Software

Autori: Andersson, Måns I.; Murugan, N. Arul; Podobas, Artur; Markidis, Stefano
Pubblicato in: PPAM22, Numero Lecture Notes in Computer Science, vol 13826, 2023, Pagina/e 333–345, ISBN 978-3-031-30442-2
Editore: Springer
DOI: 10.48550/arxiv.2208.13658

Lifting C semantics for dataflow optimization

Autori: Calotoiu, Alexandru; Ben-Nun, Tal; Kwasniewski, Grzegorz; Licht, Johannes de Fine; Schneider, Timo; Schaad, Philipp; Hoefler, Torsten
Pubblicato in: ICS '22: Proceedings of the 36th ACM International Conference on Supercomputing, 2022
Editore: Association for Computing Machinery
DOI: 10.1145/3524059.3532389

ecoHMEM: Improving Object Placement Methodology for Hybrid Memory Systems in HPC

Autori: Jordà Peroliu, Marc; Rai, Siddharth; Ayguadé Parra, Eduard; Labarta Mancho, Jesús José; Peña Monferrer, Antonio José
Pubblicato in: Crossref, Numero 18, 2022
Editore: IEEE
DOI: 10.1109/cluster51413.2022.00040

PROGRAML: A Graph-based Program Representation for Data Flow Analysis and Compiler Optimizations

Autori: Chris Cummins and Zacharias V. Fisches and Tal Ben-Nun and Torsten Hoefler and Michael O’Boyle and Hugh Leather
Pubblicato in: ICML'21, 2022
Editore: -

Combining Dynamic Concurrency Throttling with Voltage and Frequency Scaling on Task-based Programming Models

Autori: Antoni Navarro Muñoz; Arthur F. Lorenzon; Eduard Ayguadé Parra; Vicenç Beltran Querol
Pubblicato in: ICPP, 2021
Editore: Association for Computing Machinery
DOI: 10.1145/3472456.3472471

Advanced synchronization techniques for task-based runtime systems.

Autori: D. Álvarez, K. Sala, M. Maroñas, A. Roca, V. Beltran
Pubblicato in: Proceedings of PPoPP 2021, 2021
Editore: ACM
DOI: 10.1145/3437801.3441601

FMI: Fast and Cheap Message Passing for Serverless Functions

Autori: Copik, Marcin; Böhringer, Roman; Calotoiu, Alexandru; Hoefler, Torsten
Pubblicato in: Proceedings of the 37th International Conference on Supercomputing, Numero 18, 2023
Editore: ACM
DOI: 10.1145/3577193.3593718

Deinsum: Practically I/O Optimal Multilinear Algebra

Autori: A. Nikolaos Ziogas, G. Kwasniewski, T. Ben-Nun, T. Schneider, T. Hoefler
Pubblicato in: 2022
Editore: SC'22
DOI: 10.48550/arxiv.2206.08301

a benchmarking suite for high-performance NumPy

Autori: Alexandros Nikolaos Ziogas; Tal Ben-Nun; Timo Schneider; Torsten Hoefler
Pubblicato in: Proceedings of the ISC 2021, 2021
Editore: Association for Computing Machinery
DOI: 10.1145/3447818.3460360

Influence of Network Performance Variability on Application Scalability

Autori: Daniele De Sensi; Tiziano De Matteis; Konstantin Taranov; Salvatore Di Girolamo; Tobias Rahn; Torsten Hoefler
Pubblicato in: Proceedings of the ACM on Measurement and Analysis of Computing Systems, 6 (3), Numero 2, 2022
Editore: ACM
DOI: 10.1145/3570609

A framework for hierarchical single-copy MPI collectives on multicore nodes

Autori: G. Katevenis-Bitzos, M. Ploumidis, and M. Marazakis
Pubblicato in: IEEE Cluster 2022, Numero Presented at conference, 2022
Editore: IEEE

User-guided Page Merging for Memory Deduplication in Serverless Systems

Autori: Qiu, Wei; Copik, Marcin; Wang, Yun; Calotoiu, Alexandru; Hoefler, Torsten
Pubblicato in: 2023 IEEE International Conference on Big Data (Big Data), Numero 18, 2023
Editore: IEEE
DOI: 10.1109/bigdata59044.2023.10386487

Classification of Solar Flares using Data Analysis and Clustering of Active Regions

Autori: Hanne Baeke; Jorge Amaya; Giovanni Lapenta
Pubblicato in: Crossref, Numero 1, 2023
Editore: ESS Open Archive
DOI: 10.22541/essoar.167336864.46114556/v1

Processing in Memory: The Tipping Point

Autori: Petar Radojković, Paul Carpenter, Pouya Esmaili-Dokht, Rémy Cimadomo, Henri-Pierre Charles, Abu Sebastian, Paolo Amato
Pubblicato in: ETP4HPC White Paper, 2021
Editore: ETP4HPC
DOI: 10.5281/zenodo.4767489

FTIO: Detecting I/O Periodicity Using Frequency Techniques

Autori: Tarraf, Ahmad; Bandet, Alexis; Zanon Boito, Francieli; Pallez, Guillaume; Wolf, Felix
Pubblicato in: https://inria.hal.science/hal-04382142, Numero 2, 2023
Editore: arxiv
DOI: 10.48550/arxiv.2306.08601

Modular Supercomputing Architecture

Autori: Suarez, Estela; Eicker, Norbert; Moschny, Thomas; Pickartz, Simon; Clauss, Carsten; Plugaru, Valentin; Herten, Andreas; Michielsen, Kristel; Lippert, Thomas
Pubblicato in: ETP4HPC White Papers, 2022
Editore: ETP4HPC
DOI: 10.5281/zenodo.6508394

Heterogeneous High Performance Computing

Autori: P. Carpenter, U.-U. Haus, E. Laure, S. Narasimhamurthy, E. Suarez
Pubblicato in: ETP4HPC White Paper, 2022
Editore: ETP4HPC
DOI: 10.5281/zenodo.6090425

Task-Based Performance Portability in HPC

Autori: Olivier Aumage, Paul Carpenter, Siegfried Benkner
Pubblicato in: ETP4HPC White Paper, 2021
Editore: ETP4HPC

HPC for Urgent Decision-Making

Autori: M. Marazakis, M.Duranton, D. Pleiter, G. Taffoni, and H.C. Hoppe
Pubblicato in: ETP4HPC White Paper, 2022
Editore: ETP4HPC
DOI: 10.5281/zenodo.6107362

Critical Analysis of the Modular Supercomputing Architecture

Autori: E. Suarez, N. Eicker, Th.Moschny, Th. Lippert
Pubblicato in: Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project, 2021
Editore: FZJ Zentralbibliothek Verlag

An OpenMP free agent threads implementation

Autori: Lopez, Victor; Criado, Joel; Peñacoba, Raúl; Ferrer, Roger; Teruel, Xavier; Garcia-Gasulla, Marta
Pubblicato in: Lecture Notes in Computer Science - OpenMP: Enabling Massive Node-Level Parallelism, 2021
Editore: Springer
DOI: 10.1007/978-3-030-85262-7_15

Best practices guide

Autori: A. Kreuzer, J. Kreutz, B. Steinbusch
Pubblicato in: Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project, 2021
Editore: FZJ Zentralbibliothek Verlag

Space weather with DLMOS, xPic and GMM

Autori: J.Amaya
Pubblicato in: Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project, 2021
Editore: FZJ Zentralbibliothek Verlag

The DEEP-EST project

Autori: E. Suarez, A.Kreuzer, N. Eicker, Th. Lippert
Pubblicato in: Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project, 2021
Editore: FZJ Zentralbibliothek Verlag

Simulating Structural Plasticity of the Brain more Scalable than Expected

Autori: F. Czappa, A. Geiß, F. Wolf
Pubblicato in: Journal of Parallel and Distributed Computing, Numero 07437315, 2022, ISSN 0743-7315
Editore: Academic Press
DOI: 10.1016/j.jpdc.2022.09.001

Mitigating the NUMA Effect on Task-Based Runtime Systems

Autori: M. Maroñas, A. Navarro, E. Ayguadé and V. Beltran
Pubblicato in: The Journal of Supercomputing, Numero 4, 2023, ISSN 0920-8542
Editore: Kluwer Academic Publishers
DOI: 10.1007/s11227-023-05164-9

Operational Data Analytics in practice: Experiences from design to deployment in production HPC environments

Autori: Alessio Netti, Michael Ott, Carla Guillen, Daniele Tafani, Martin Schulz
Pubblicato in: Parallel Computing (ParCo), Numero 01678191, 2022, ISSN 0167-8191
Editore: Elsevier BV
DOI: 10.1016/j.parco.2022.102950

Towards leveraging collective performance with the support of MPI 4.0 features in MPC

Autori: S. Bouhrour, T. Pepin, J. Jaeger
Pubblicato in: Journal on Parallel Computing (ParCo), Numero Volume 109, March 2022, 102860, 2022, ISSN 0167-8191
Editore: Elsevier BV
DOI: 10.1016/j.parco.2021.102860

Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project

Autori: A. Kreuzer, E. Suarez, N. Eicker, Th. Lippert
Pubblicato in: Porting applications to a Modular Supercomputer - Experiences from the DEEP-EST project, 2021
Editore: FZJ Zentralbibliothek Verlag

È in corso la ricerca di dati su OpenAIRE...

DEEP – SOFTWARE FOR EXASCALE ARCHITECTURES

Risultati finali

Pubblicazioni

Condividi questa pagina

Scarica