Modeling parallel applications performance on heterogeneous systems

Al-Jaroodi, Jameela; Mohamed, Nader; Jiang, Hong; Swanson, David

doi:10.1109/ipdps.2003.1213298

Cited by 17 publications

(13 citation statements)

References 9 publications

(11 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this approach, an analytical model that parametrizes the architecture and the algorithm is instantiated for each specific case so as to optimize program execution. This strategy is considerably more general than the previous one, though more difficult to apply since the modeling process is not trivial [10,11], nor is its subsequent instantiation and minimization for each case. A search of the literature yields some generic tools such as mpC [12,13] and HeteroMPI [14,15] which provide the mechanisms that allow algorithms to be adapted to heterogeneous architectures, but which also require more input from the user and are more code intrusive.…”

Section: Background and Objectivesmentioning

confidence: 99%

Dynamic Load Balancing on Dedicated Heterogeneous Systems

Galindo¹,

Almeida²,

Badía³

2008

Recent Advances in Parallel Virtual Machine and Message Passing Interface

View full text Add to dashboard Cite

Abstract. Parallel computing in heterogeneous environments is drawing considerable attention due to the growing number of these kind of systems. Adapting existing code and libraries to such systems is a fundamental problem. The performance of this code is affected by the large interdependence between the code and these parallel architectures. We have developed a dynamic load balancing library that allows parallel code to be adapted to heterogeneous systems for a wide variety of problems. The overhead introduced by our system is minimal and the cost to the programmer negligible. The strategy was validated on several problems to confirm the soundness of our proposal.

show abstract

Section: Background and Objectivesmentioning

confidence: 99%

Dynamic Load Balancing on Dedicated Heterogeneous Systems

Galindo¹,

Almeida²,

Badía³

2008

Recent Advances in Parallel Virtual Machine and Message Passing Interface

View full text Add to dashboard Cite

show abstract

“…The average service time and the service reliability of a heterogeneous, five-server DCS have also been optimized by solving the optimization problems (13) to (15). It has been assumed that the parallel application is partitioned in M ¼ 250 tasks.…”

Section: Application Of the Theory To Optimal Task Reallocation In Dcssmentioning

confidence: 99%

“…3 and 4 , the maximal service reliability is achieved when, respectively, 27 and 10 tasks are transferred from server 1 to 2. However, the exponential approximation Table 2, we have listed results for the average service time and the service reliability of the five-server DCS, under severe network-delay conditions, when the exact and the approximated model have been employed to solve for the optimization problems stated in (13) and (15). The stochastic model evaluated in this example is Pareto with finite variance ("Pareto 1" model).…”

Section: Application Of the Theory To Optimal Task Reallocation In Dcssmentioning

confidence: 99%

“…The time measurements have been obtained using a standard nondedicated desktop computer, equipped with a 2.20 GHz dual core processor, 8 GBytes of RAM and running a 64 bits, Linux-based operating system. Equations and algorithms to solve the optimization problems (13) and (15) were coded in C, and the GNU scientific library was exploited in order to efficiently calculate the integrals of each equation. The average computing times reported in Table 2 correspond to averaging ten executions of the software that solves the optimization problems.…”

Section: Application Of the Theory To Optimal Task Reallocation In Dcssmentioning

confidence: 99%

“…Such model is obtained under the assumption that all the timedevents that collectively govern the behavior of the DCS (e.g., task-execution times, communication arrival times, loadtransfer times and server failure times) follow exponential distributions [12], [15], [16], [17], [18], [19]. The vast amount of research on Markovian models has yielded several modeling and analysis tools such as TimeNET and WebSPN [12], [17], [18], [19].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Performance and Reliability of Non-Markovian Heterogeneous Distributed Computing Systems

Pezoa

Hayat

2012

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Abstract-Average service time, quality-of-service (QoS), and service reliability associated with heterogeneous parallel and distributed computing systems (DCSs) are analytically characterized in a realistic setting for which tangible, stochastic communication delays are present with nonexponential distributions. The departure from the traditionally assumed exponential distributions for event times, such as task-execution times, communication arrival times and load-transfer delays, gives rise to a non-Markovian dynamical problem for which a novel age dependent, renewal-based distributed queuing model is developed. Numerical examples offered by the model shed light on the operational and system settings for which the Markovian setting, resulting from employing an exponentialdistribution assumption on the event times, yields inaccurate predictions. A key benefit of the model is that it offers a rigorous framework for devising optimal dynamic task reallocation (DTR) policies systematically in heterogeneous DCSs by optimally selecting the fraction of the excess loads that need to be exchanged among the servers, thereby controlling the degree of cooperative processing in a DCSs. Key results on performance prediction and optimization of DCSs are validated using Monte-Carlo (MC) simulation as well as experiments on a distributed computing testbed. The scalability, in the number of servers, of the age-dependent model is studied and a linearly scalable analytical approximation is derived.

show abstract