Filters








29,394 Hits in 4.6 sec

GLTO: On the Adequacy of Lightweight Thread Approaches for OpenMP Implementations

Adrian Castello, Sangmin Seo, Rafael Mayo, Pavan Balaji, Enrique S. Quintana-Orti, Antonio J. Pena
2017 2017 46th International Conference on Parallel Processing (ICPP)  
We have developed GLTO, an OpenMP implementation over the recentlyemerged Generic Lightweight Threads (GLT) API.  ...  In this paper we use GLTO to analyze different scenarios where OpenMP implementations may benefit from the use of either LWT or pthreads.  ...  Despite this API does not support the complete set of the native LWT semantics, the selected group is sufficient to implement our OpenMP runtime.  ... 
doi:10.1109/icpp.2017.15 dblp:conf/icpp/CastelloSMBQP17 fatcat:varshzcssbeq3mlytl5lqctiiq

On the adequacy of lightweight thread approaches for high-level parallel programming models

Adrián Castelló, Rafael Mayo, Kevin Sala, Vicenç Beltran, Pavan Balaji, Antonio J. Peña
2018 Future generations computer systems  
The most popular PMs, such as OpenMP or OmpSs, are directive-based: the complexity of the hardware is hidden by the underlying runtime system, improving coding productivity.  ...  Our work reveals those scenarios where LWTs overperform pthread-based solutions and compares the performance between an ad hoc solution and a generic implementation.  ...  If we compare the GLTO options among them, we observe the effect of different implementation details of the underlying libraries.  ... 
doi:10.1016/j.future.2018.02.016 fatcat:pbo2kyo4sjgzppbxjo2cf7ofza

A Taxonomy Of Task-Based Parallel Programming Technologies For High-Performance Computing

Peter Thoman, Kiril Dichev, Khalid Hasanov, Roman Iakymchuk, Xavier Aguilar, Thomas Heller, Philipp Gschwandtner, Pierre Lemarinier, Stefano Markidis, Herbert Jordan, Thomas Fahringer, Kostas Katrinis (+2 others)
2017 Zenodo  
Unfortunately, despite the fact that dozens of different task-based systems exist today and are actively used for parallel and high-performance computing (HPC), no comprehensive overview or classification  ...  In this paper, we provide an initial task-focused taxonomy for HPC technologies, which covers both programming interfaces and runtime mechanisms.  ...  In this section, we provide a taxonomy of many-task runtime systems, which is illustrated in Figure 2 . A crucial difference among various many-task runtime systems is their target architecture.  ... 
doi:10.5281/zenodo.1119094 fatcat:kbuhio5hu5bs7kqkuj5s4jijdi

EngineCL: Usability and Performance in Heterogeneous Computing [article]

Raúl Nozal, Jose Luis Bosque, Ramón Beivide (Universidad de Cantabria)
2018 arXiv   pre-print
This paper presents EngineCL, a new OpenCL-based runtime system that notably simplifies the execution of a single massive data-parallel kernel on a heterogeneous system.  ...  However, due to their heterogeneity they are very complex to program and even more to achieve performance portability on different devices.  ...  He is also responsible for selecting the devices on which each kernel is going to be enqueued as well as for partitioning the data among them.  ... 
arXiv:1805.02755v1 fatcat:gbnozap5qfakfhi3g27pjf77zi

A taxonomy of task-based parallel programming technologies for high-performance computing

Peter Thoman, Kiril Dichev, Thomas Heller, Roman Iakymchuk, Xavier Aguilar, Khalid Hasanov, Philipp Gschwandtner, Pierre Lemarinier, Stefano Markidis, Herbert Jordan, Thomas Fahringer, Kostas Katrinis (+2 others)
2018 Journal of Supercomputing  
Unfortunately, despite the fact that dozens of different task-based systems exist today and are actively used for parallel and high-performance computing (HPC), no comprehensive overview or classification  ...  In this paper, we provide an initial task-focused taxonomy for HPC technologies, which covers both programming interfaces and runtime mechanisms. We  ...  In this section, we provide a taxonomy of many-task runtime systems, which is illustrated in Fig. 2 . A crucial difference among various many-task runtime systems is their target architecture.  ... 
doi:10.1007/s11227-018-2238-4 fatcat:fctzmtp3n5fithxfchl5rub7j4

Ensuring interoperable service-oriented systems through engineered self-healing

Giovanni Denaro, Mauro Pezzè, Davide Tosi
2009 Proceedings of the 7th joint meeting of the European software engineering conference and the ACM SIGSOFT symposium on The foundations of software engineering on European software engineering conference and foundations of software engineering symposium - ESEC/FSE '09  
For instance, problems that derive from behavior incompatibility are quite frequent in service-oriented applications that dynamically bind service implementations that match API specifications.  ...  The runtime infrastructure enacts the configured self-healing strategies.  ...  to select, among the potential mismatches identified for an API specification, the subset that is relevant for a client.  ... 
doi:10.1145/1595696.1595734 dblp:conf/sigsoft/DenaroPT09 fatcat:7b6snzs2nbbcrhmwdujkh2y46a

How Do API Selections Affect the Runtime Performance of Data Analytics Tasks?

Yida Tao, Shan Tang, Yepang Liu, Zhiwu Xu, Shengchao Qin
2019 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE)  
In this paper, we propose a novel approach to extracting alternative implementations that invoke different data analytics APIs to solve the same tasks.  ...  We observed that developers sometimes use alternative data analytics APIs to improve program runtime performance while preserving functional equivalence.  ...  The results show that alternative implementations using different data analytics APIs do improve task runtime performance, and sometimes the improvement is quite significant.  ... 
doi:10.1109/ase.2019.00067 dblp:conf/kbse/TaoTLXQ19 fatcat:wxwzujzirfhajgsbadhiwkjqme

A Framework for Performance-Aware Composition of Applications for GPU-Based Systems

Usman Dastgeer, Christoph Kessler
2013 2013 42nd International Conference on Parallel Processing  
as well as cases when they are learned at runtime.  ...  We demonstrate several key features of our framework relating to performanceaware composition including implementation selection, both with performance characteristics being known (or learned) beforehand  ...  In the sequential homogeneous computing paradigm, implementation selection normally considers which implementation is (expected to be) the fastest one for a certain input among all implementation variants  ... 
doi:10.1109/icpp.2013.83 dblp:conf/icpp/DastgeerK13 fatcat:3vpquv7kmbejbpdtzpysnblg64

Gdev: First-Class GPU Resource Management in the Operating System

Shinpei Kato, Michael McThrow, Carlos Maltzahn, Scott A. Brandt
2012 USENIX Annual Technical Conference  
Gdev further provides a GPU scheduling scheme to virtualize a physical GPU into multiple logical GPUs, enhancing isolation among working sets of multi-tasking systems.  ...  Our evaluation conducted on Linux and the NVIDIA GPU shows that the basic performance of our prototype implementation is reliable even compared to proprietary software.  ...  The Gdev API is implemented in this runtime library. The kernel symbols of the API functions are exported so that other OS modules can call them.  ... 
dblp:conf/usenix/KatoMMB12 fatcat:6mvayfpry5fp5ldq7oimfzrmmi

Provisioning Software-Defined IoT Cloud Systems

Stefan Nastic, Sanjin Sehic, Duc-Hung Le, Hong-Linh Truong, Schahram Dustdar
2014 2014 International Conference on Future Internet of Things and Cloud  
In this paper, we introduce the concept of software-defined IoT units -a novel approach to IoT cloud computing that encapsulates fine-grained IoT resources and IoT capabilities in well-defined APIs in  ...  mainly due to the complexity involved with provisioning large-scale IoT cloud systems and diversity of their requirements in terms of IoT resources consumption, customization of IoT capabilities and runtime  ...  Because we treat the functional and configuration units in a similar manner (see Section IV-B), configuration models can be stored, reused, modified at runtime and even shared among different stakeholders  ... 
doi:10.1109/ficloud.2014.52 dblp:conf/ficloud/NasticSLTD14 fatcat:sfysl4dmyrd3hda4mycot2zk7u

Active Harmony: Towards Automated Performance Tuning

C. Tapus, I-Hsin Chung, J.K. Hollingsworth
2002 ACM/IEEE SC 2002 Conference (SC'02)  
We present the Library Specification Layer which helps program library developers expose multiple variations of the same API using different algorithms.  ...  The Library Specification Language helps to select the most appropriate program library to tune the overall performance.  ...  It also provides the indirection to allow runtime switching among the different implementations.  ... 
doi:10.1109/sc.2002.10062 dblp:conf/sc/TapusCH02 fatcat:ahm3a4hs3zh4bp3qmi2ajrvf2a

NFVactor: A Resilient NFV System using the Distributed Actor Model [article]

Jingpu Duan, Xiaodong Yi, Shixiong Zhao, Chuan Wu, Heming Cui, Franck Le
2018 arXiv   pre-print
Moreover, NFVactor achieves transparent resilience: once a new NF is implemented for NFVactor, the NF automatically acquires resilience support.  ...  to provide a per-flow execution environment, so that each flow can replicate and migrate itself with improved parallelism, while the efficiency of the actor model is guaranteed by a carefully designed runtime  ...  There is no need to implement the last 3 APIs in Tbl.  ... 
arXiv:1805.06160v1 fatcat:q34fneusfjc4beouejwym5qfze

ENHANCING CLOUD APPLICATION DEVOPS USING DYNAMICALLY TAILORED DEPLOYMENT ENGINES

2016 Services Transactions on Cloud Computing  
The core of our work is based on generating APIs for arbitrary deployment executables such as scripts and plans that perform different tasks in the automated deployment process.  ...  As a result, deployment tasks can be triggered through generated API endpoints, abstracting from lower-level, technical details of diverse deployment automation tooling.  ...  Microservices interact among each other through language-agnostic APIs (e.g., HTTP-based REST APIs), so each application component can potentially be implemented based on a different technology stack.  ... 
doi:10.29268/stcc.2016.4.1.2 fatcat:pcoiyesc2zejdmxqojfuitkszq

Comprehending implementation recipes of framework-provided concepts through dynamic analysis

Abbas Heydarnoori, Krzysztof Czarnecki
2007 Companion to the 22nd ACM SIGPLAN conference on Object oriented programming systems and applications companion - OOPSLA '07  
Application developers often use example applications as a guide to learn how to implement a framework-provided concept.  ...  FUDA integrates a new dynamic slicing approach with clustering and data mining techniques to generate the implementation recipes of a desired concept.  ...  An AIDG is a directed graph that represents the potential runtime data dependencies among the events in the framework API interaction trace with respect to their order of execution.  ... 
doi:10.1145/1297846.1297904 dblp:conf/oopsla/HeydarnooriC07 fatcat:fo2pavz755cizp3dh7cidq4ypm

Dyn Tail - Dynamically Tailored Deployment Engines for Cloud Applications

Johannes Wettinger, Uwe Breitenbucher, Frank Leymann
2015 2015 IEEE 8th International Conference on Cloud Computing  
Microservices interact among each other through language-agnostic APIs (e.g., HTTP-based REST APIs), so each application component can potentially be implemented based on a different technology stack.  ...  All three phases (design time, build time, and runtime) are linked among each other, i.e., feedback loops are supported to go from the runtime phase back to the build time and design time phase in order  ... 
doi:10.1109/cloud.2015.63 dblp:conf/IEEEcloud/WettingerBL15 fatcat:km4uhkv7cvdg7fkc4anfbqt57u
« Previous Showing results 1 — 15 out of 29,394 results