527 Hits in 4.3 sec

PaRSEC: Exploiting Heterogeneity to Enhance Scalability

George Bosilca, Aurelien Bouteiller, Anthony Danalis, Mathieu Faverge, Thomas Herault, Jack J. Dongarra
2013 Computing in science & engineering (Print)  
Having reached a hard upper limit on clock frequencies, designers began to seek higher performance by increasing the number of computing resources on each chip; the manycore revolution began.  ...  These new approaches must react and adapt quickly to unexpected contentions and delays, and they must provide the execution environment with sufficient intelligence and flexibility to rearrange the execution  ...  The PaRSEC Runtime PaRSEC employs the dataflow programming and execution model to provide a dynamic platform that can address the challenges posed by distributed heterogeneous hardware resources.  ... 
doi:10.1109/mcse.2013.98 fatcat:qyghrkdwyjhs5porciopjjipi4

Challenges of Translating HPC Codes to Workflows for Heterogeneous and Dynamic Environments

Fayssal Benkhaldoun, Christophe Cerin, Imad Kissami, Walid Saad
2017 2017 International Conference on High Performance Computing & Simulation (HPCS)  
We show how to move to the field of "HPC as a Service" in order to use heterogeneous platforms.  ...  We discuss in particular of the impact on the workflow engine of such dynamic feature.  ...  The object oriented ADAPT platform is able to do coupling between finite volumes and finite elements methods.  ... 
doi:10.1109/hpcs.2017.130 dblp:conf/ieeehpcs/BenkhaldounCKS17 fatcat:cxr7q7s46jg4zbf4etyky6llue

MB++: An Integrated Architecture for Pervasive Computing and High-Performance Computing

David J. Lillethun, David Hilley, Seth Horrigan, Umakishore Ramachandran
2007 13th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA 2007)  
The transformation engine executes dataflow graphs of transformations on high-performance computing resources.  ...  We have implemented the architecture and show performance results that demonstrate that our implementation scales well with increasing workload, commensurate with the available HPC resources.  ...  The TE also provide platform independence to transformers, since transformer developers may not be able to predict the platform on which the transformer may be run.  ... 
doi:10.1109/rtcsa.2007.47 dblp:conf/rtcsa/LillethunHHR07 fatcat:oyjfedl4xneftj7gd7aziw3cnq

Editorial for the special issue on operating systems and programming systems for HPC

Xiaobing Feng, Minyi Guo
2020 CCF Transactions on High Performance Computing  
Related compiling and runtime optimizations are also presented, and evaluation on two supercomputing platforms shows its performance superiority to existing programming models. • The second paper written  ...  The second part of the special issue, consisting of two research papers, focuses on the runtime system and operating system support for HPC applications. Customized * Xiaobing Feng  ...  Targeting a heterogeneous architecture with both dataflow and control-flow execution models, a novel runtime system is designed to adaptively offload each subgraph to an appropriate processor, accompanied  ... 
doi:10.1007/s42514-020-00053-6 fatcat:nthaiyn6m5eqvisdxwiz7r7u2m

A Survey: Runtime Software Systems for High Performance Computing

2017 Supercomputing Frontiers and Innovations  
Runtimes provide adaptive means to reduce the effects of starvation, latency, overhead, and contention.  ...  execution to guide task scheduling and resource management for dynamic adaptive control.  ...  With perhaps one important exception, very limited use of runtime systems has been made in the arena of HPC.  ... 
doi:10.14529/jsfi170103 fatcat:yqj65kpvhngovcmgrr46vwwr6i


Adrian Cristal, Sergi Madonar, Miquel Pericàs, Pedro Trancoso, Micha vor dem Berge, Gunnar Billung-Meyer, Stefan Krupop, Wolfgang Christmann, Frank Klawonn, Amani Mihklafi, Tobias Becker, Osman S. Unsal (+24 others)
2018 Proceedings of the 15th ACM International Conference on Computing Frontiers - CF '18  
The aim is to attain one order of magnitude energy savings from the edge to the converged cloud/HPC.  ...  The LEGaTO project will leverage task-based programming models to provide a software ecosystem for Madein-Europe heterogeneous hardware composed of CPUs, GPUs, FPGAs and dataflow engines.  ...  runtime execution model is well adapted for streaming accelerators such as FPGAs or DFEs.  ... 
doi:10.1145/3203217.3205339 dblp:conf/cf/CristalUMCCBJAS18 fatcat:zhakz5fi55fb5dqmxtpjols6ta

SciFlow: A dataflow-driven model architecture for scientific computing using Hadoop

Pengfei Xuan, Yueli Zheng, Sapna Sarupria, Amy Apon
2013 2013 IEEE International Conference on Big Data  
on a Hadoop platform.  ...  It provides an efficient mechanism for building a parallel scientific application with dataflow patterns, and enables the design, deployment, and execution of data intensive, many-task computing tasks  ...  ACKNOWLEDGMENTS This research is sponsored in part by NSF awards CNS-1228312 and OCI-1212680, and used HPC resources of the Clemson Computing and Information Technology.  ... 
doi:10.1109/bigdata.2013.6691725 dblp:conf/bigdataconf/XuanZSA13 fatcat:7sptcg6ge5cxbcemhfomxkbyli

SAVE: Towards Efficient Resource Management in Heterogeneous System Architectures [chapter]

G. Durelli, M. Coppola, K. Djafarian, G. Kornaros, A. Miele, M. Paolino, Oliver Pell, Christian Plessl, M. D. Santambrogio, C. Bolchini
2014 Lecture Notes in Computer Science  
The SAVE project will develop HW/SW/OS components that allow for deciding at runtime the mapping of the computation kernels on the appropriate type of resource, based on the current system context and  ...  Self-adaptiveness and hardware-assisted virtualization are the two key-enabling technologies for this kind of architectures, to allow the efficient exploitation of the available resources based on the  ...  One instance of this approach that is currently receiving a lot of attention in high-performance computing (HPC) are dataflow engines (DFEs), that implement custom, high-throughput processing pipelines  ... 
doi:10.1007/978-3-319-05960-0_38 fatcat:txyxyy3oxfgxxlejazrypktzti

CAOS: CAD as an Adaptive Open-Platform Service for High Performance Reconfigurable Systems [chapter]

Marco Rabozzi
2019 SpringerBriefs in Applied Sciences and Technology  
In this chapter, we present CAD as an Adaptive Open-platform Service (CAOS), a platform to guide the application developer in the implementation of efficient hardware-software solutions for high performance  ...  The platform assists the designer from the high-level analysis of the code, towards the optimization and implementation of the functionalities to be accelerated on the reconfigurable nodes.  ...  CAOS: CAD as an Adaptive Open-Platform Service for High Performance … CAOS: CAD as an Adaptive Open-Platform Service for High Performance … CAOS: CAD as an Adaptive Open-Platform Service for  ... 
doi:10.1007/978-3-030-32094-2_8 fatcat:zpang4mxhbarnacnihbujgjsya

TensorFlow Doing HPC [article]

Steven W. D. Chien, Stefano Markidis, Vyacheslav Olshevsky, Yaroslav Bulatov, Erwin Laure, Jeffrey S. Vetter
2019 arXiv   pre-print
Running our TensorFlow STREAM benchmark, we obtain over 50% of theoretical communication bandwidth on our testing platform.  ...  However, very few experiments have been conducted to evaluate TensorFlow performance when running HPC workloads on supercomputers.  ...  Experiments were performed on resources provided by the Swedish National Infrastructure for Computing (SNIC) at PDC Center for High Performance Computing and HPC2N.  ... 
arXiv:1903.04364v1 fatcat:wxdhhe4ewncqjijzqsctkyuvkq

Using data-flow analysis in MAS for power-aware HPC runs

Sebastien Varrette, Gregoire Danoy, Mateusz Guzek, Pascal Bouvry
2013 2013 International Conference on High Performance Computing & Simulation (HPCS)  
Tasks in G and therefore E are computed on a distributed computing platform, typically an HPC system.  ...  Coupled with a holistic modelling of the HPC platform introduced in a previous article [5] and adapted to the latest hardware advances, work-stealing operations and local decisions as regards the power  ... 
doi:10.1109/hpcsim.2013.6641407 dblp:conf/ieeehpcs/VarretteDGB13 fatcat:q4vagt4o4nazdobx5pml7rlm7a

ParaMASK: A Multi-Agent System for the efficient and dynamic adaptation of HPC workloads

Mateusz Guzek, Xavier Besseron, Sebastien Varrette, Gregoire Danoy, Pascal Bouvry
2014 2014 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT)  
In this context, we describe here a dynamic and flexible adaptation scheme based on a Multi-Agent System (MAS) to handle parallel or distributed executions in an HPC environment.  ...  The growing parallelism and heterogeneity of modern computing infrastructures such as High Performance Computing (HPC) platforms raises new challenges to their programmers and users.  ...  It is of special interest for new HPC platforms, where runtime fine-tuning is dependent on multiple changing parameters, to optimize the system dynamically.  ... 
doi:10.1109/isspit.2014.7300600 dblp:conf/isspit/GuzekBVDB14 fatcat:o6og37bkqbg7jpvawckvkgbezq

EXTRA: Towards an Efficient Open Platform for Reconfigurable High Performance Computing

Catalin Bogdan Ciobanu, Ana Lucia Varbanescu, Dionisios Pnevmatikatos, George Charitopoulos, Xinyu Niu, Wayne Luk, Marco D. Santambrogio, Donatella Sciuto, Muhammed Al Kadi, Michael Huebner, Tobias Becker, Georgi Gaydadjiev (+5 others)
2015 2015 IEEE 18th International Conference on Computational Science and Engineering  
feature instead of an add-on.  ...  EXTRA covers the entire stack from architecture up to the application, focusing on the fundamental building blocks for run-time reconfigurable exascale HPC systems: new chip architectures with very low  ...  be discovered at runtime.  ... 
doi:10.1109/cse.2015.54 dblp:conf/cse/CiobanuVPCNLSSK15 fatcat:h5seovvsxzdcbpahv72p67p6qy

Keeping track of user steering actions in dynamic workflows

Renan Souza, Vítor Silva, Jose J. Camata, Alvaro L.G.A. Coutinho, Patrick Valduriez, Marta Mattoso
2019 Future generations computer systems  
In addition to losing track of changes, one misses opportunities to learn from the adaptation data (i.e., data generated when humans adapt a certain dataflow element) with the associated dataflow.  ...  Moreover, opportunities to use the data for AI-based assistants recommending on what to adapt next, based on a database of adaptations, are lost.  ...  Simulations that use highly parallel libraries or adaptive parallel algorithms, already implement parallel execution control and scheduling on HPC machines.  ... 
doi:10.1016/j.future.2019.05.011 fatcat:hppilac4vzbl7mgwqdgv3t27ca

The Center for Plasma Edge Simulation Workflow Requirements

S.A. Klasky, B. Ludaescher, M. Parashar
2006 22nd International Conference on Data Engineering Workshops (ICDEW'06)  
Our approach is to couple these codes via a scientific workflow system, Kepler-HPC.  ...  Kepler-HPC will enhance Kepler with capabilities such as code coupling and data redistribution, high volume data transfers and interactive (and autonomic) monitoring, steering and debugging, which will  ...  This constraint is due to the fact that the codes running on the separate platforms must be "in-synch" with one another.  ... 
doi:10.1109/icdew.2006.143 dblp:conf/icde/KlaskyLP06 fatcat:cnr3je2vqzhvdopumtlfdl3lia
« Previous Showing results 1 — 15 out of 527 results