Dynamic Malleability in Iterative MPI Applications

Maghraoui, Kaoutar El; Desell, Travis; Szymanski, Boleslaw K.; Varela, Carlos A.

doi:10.1109/ccgrid.2007.45

Cited by 45 publications

(30 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Process Checkpointing and Migration (PCM) [18,19] is a runtime system built in the context of the Internet Operating System (IOS) [20] and uses process migration to provide malleability to MPI applications. The PCM/IOS library allows MPI programs to reconfigure themselves to adapt to the available processors as well as the performance of the application by using either split/merge operations or process migration.…”

Section: Related Workmentioning

confidence: 99%

Enhancing the performance of malleable MPI applications by using performance-aware dynamic reconfiguration

et al. 2015

View full text Add to dashboard Cite

The work in this paper focuses on providing malleability to MPI applications by using a novel performance-aware dynamic reconfiguration technique. This paper describes the design and implementation of Flex-MPI, an MPI library extension which can automatically monitor and predict the performance of applications, balance and redistribute the workload, and reconfigure the application at runtime by changing the number of processes. Unlike existent approaches, our reconfiguring policy is guided by user-defined performance criteria. We focus on iterative SPMD programs, a class of applications with critical mass within the scientific community. Extensive experiments show that Flex-MPI can improve the performance, parallel efficiency, and cost-efficiency of MPI programs with a minimal effort from the programmer.

show abstract

Section: Related Workmentioning

confidence: 99%

Enhancing the performance of malleable MPI applications by using performance-aware dynamic reconfiguration

et al. 2015

View full text Add to dashboard Cite

show abstract

“…With this motivation, this paper proposes, implements and evaluates DMI (Distributed Memory Interface), which is a global-view-based PGAS (Partitioned Global Address Space) framework that enables easy programming of reconfigurable and high-performance parallel iterative computations. These parallel iterative computations include many important scientific applications such as a finite element method, a multigrid method, a particle method and iterative graph search, which have high requirement for the dynamic reconfiguration because these applications are often long-running [19].…”

Section: Backgrounds and Goalsmentioning

confidence: 99%

“…We also confirmed that DMI has higher programmability than MPI for those irregular scientific applications thanks to the global address space (Section 5). To the best of our knowledge, this is the first work that achieves the reconfiguration based on a PGAS model, although there have been several works based on a message passing model such as MPI [8], [19], [20], [26].…”

Section: Contributionsmentioning

confidence: 99%

“…In order to run his computation efficiently on the unpredicted number of resources, the programmer should create very many processes for good load balancing. However, not only is how many processes the programmer should initially create often unclear in a nonartificial dynamic environment, but also creating too many processes degrades performance dramatically [19]. The first reason for this performance degradation is the overhead of a parallel algorithm.…”

Section: Programming Models Suitable For Reconfiguration 21 Relationmentioning

confidence: 99%

“…Note that the execution time of each iteration is quite critical for the total execution time of long-running iterative applications. Second, in a processor non-virtualization model, for example adopted in SRS [26], DyRecT [8], DRMS [17] and PCM [19], [20], processes are created or killed at every reconfiguration so that one process is always assigned to one core. The advantage of this model is good performance.…”

Section: Programming Models Suitable For Reconfiguration 21 Relationmentioning

confidence: 99%

See 2 more Smart Citations

Parallel Computational Reconfiguration Based on a PGAS Model

Hara

Taura

2012

Journal of Information Processing

View full text Add to dashboard Cite

Abstract:In order to improve the resource utilization of clusters and supercomputers and thus deliver application results to users faster, it is essential for a job scheduler to expand and shrink parallel computations flexibly. In order to enable the flexible job scheduling, the parallel computations have to be reconfigurable. With this motivation, this paper proposes, implements and evaluates DMI, a global-view-based PGAS framework that enables easy programming of reconfigurable and high-performance parallel iterative computations. DMI provides programming interfaces with which a programmer can program the reconfiguration easily with a global-view. Our performance evaluations showed that DMI can efficiently adapt the parallelism of long-running parallel iterative computations, such as a real-world finite element method and large-scale iterative graph search, to the dynamic increase and decrease of available resources through the reconfiguration.

show abstract

Reparallelization techniques for migrating OpenMP codes in computational grids

Klemm

Bezold

Gabriel

et al. 2008

Concurrency and Computation

View full text Add to dashboard Cite

SUMMARYTypical computational grid users target only a single cluster and have to estimate the runtime of their jobs. Job schedulers prefer short-running jobs to maintain a high system utilization. If the user underestimates the runtime, premature termination causes computation loss; overestimation is penalized by long queue times. As a solution, we present an automatic reparallelization and migration of OpenMP applications. A reparallelization is dynamically computed for an OpenMP work distribution when the number of CPUs changes. The application can be migrated between clusters when an allocated time slice is exceeded. Migration is based on a coordinated, heterogeneous checkpointing algorithm. Both reparallelization and migration enable the user to freely use computing time at more than a single point of the grid. Our demo applications successfully adapt to the changed CPU setting and smoothly migrate between, for example, clusters in Erlangen, Germany, and Amsterdam, the Netherlands, that use different kinds and numbers of processors. Benchmarks show that reparallelization and migration impose average overheads of about 4 and 2%, respectively.

show abstract

Dynamic Malleability in Iterative MPI Applications

Cited by 45 publications

References 8 publications

Enhancing the performance of malleable MPI applications by using performance-aware dynamic reconfiguration

Enhancing the performance of malleable MPI applications by using performance-aware dynamic reconfiguration

Parallel Computational Reconfiguration Based on a PGAS Model

Reparallelization techniques for migrating OpenMP codes in computational grids

Contact Info

Product

Resources

About