Performance comparison of pure MPI vs hybrid MPI-OpenMP parallelization models on SMP clusters

Achieving high scalability with dynamically adaptive algorithms in high-performance computing (HPC) is a non-trivial task. The invasive paradigm using compute migration represents an efficient alternative to classical data migration approaches for such algorithms in HPC. We present a coredistribution scheduler which realizes the migration of computational power by distributing the cores depending on the requirements specified by one or more parallel program instances. We validate our approach with different benchmark suites for simulations with artificial workload as well as applications based on dynamically adaptive shallow water simulations, and investigate concurrently executed adaptivity parameter studies on realistic Tsunami simulations. The invasive approach results in significantly faster overall execution times and higher hardware utilization than alternative approaches. A dynamic resource management is therefore mandatory for a more efficient execution of scenarios similar to our simulations, e.g. several Tsunami simulations in urgent computing, to overcome strong scalability challenges in the area of HPC. The optimizations obtained by invasive migration of cores can be generalized to similar classes of algorithms with dynamic resource requirements.

show abstract

“…[16,27]) but little investigation has been done in combination with changing demand of resources during run time. Garcia et al's approach (c.f.…”

Section: Existing Work and Invasive Computingmentioning

confidence: 99%

“…Currently, such parallelizations are usually achieved via threading libraries such as OpenMP, via MPI, or via hybrid approaches combining distributed-and shared-memory parallel programming to obtain benefits of both implementations [16].…”

Section: Introductionmentioning

confidence: 99%

Invasive Compute Balancing for Applications with Shared and Hybrid Parallelization

Schreiber

Riesinger

Neckel

et al. 2014

Int J Parallel Prog

View full text Add to dashboard Cite

show abstract

“…Combining two programming models under a hybrid approach it is not a new proposal [22], [23], [17]. Industry has taken this approach as a method to ensure scalability, specially in high performance computing systems.…”

Section: Related Workmentioning

confidence: 99%

Speeding Up Distributed MapReduce Applications Using Hardware Accelerators

Becerra

Beltrán

Carrera

et al. 2009

2009 International Conference on Parallel Processing

View full text Add to dashboard Cite

Abstract-In an attempt to increase the performance/cost ratio, large compute clusters are becoming heterogeneous at multiple levels: from asymmetric processors, to different system architectures, operating systems and networks. Exploiting the intrinsic multi-level parallelism present in such a complex execution environment has become a challenging task using traditional parallel and distributed programming models. As a result, an increasing need for novel approaches to exploiting parallelism has arisen in these environments. MapReduce is a data-driven programming model originally proposed by Google back in 2004 as a flexible alternative to the existing models, specially devoted to hiding the complexity of both developing and running massively distributed applications in large compute clusters. In some recent works, the MapReduce model has been also used to exploit parallelism in other non-distributed environments, such as multi-cores, heterogeneous processors and GPUs. In this paper we introduce a novel approach for exploiting the heterogeneity of a Cell BE cluster linking an existing MapReduce runtime implementation for distributed clusters and one runtime to exploit the parallelism of the Cell BE nodes. The novel contribution of this work is the design and evaluation of a MapReduce execution environment that effectively exploits the parallelism existing at both the Cell BE cluster level and the heterogeneous processors level.

show abstract

“…In order to exploit two-level parallelism, OpenMP is used in conjunction with MPI, which is a parallel programming standard for shared-memory mode, and consists of a set of compiler directives and a library of support functions [25]. One of the most popular approach on the hybrid OpenMP-MPI programming is to parallelize the loop sections of the MPI application using multithreading with the aid of OpenMP.…”

Section: Implementation Of the Hybrid Parallel Fdtdmentioning

confidence: 99%

A Novel FDTD Approach Featuring Two-Level Parallelization on Pc Cluster

Liu¹,

Liang²,

Yang³

2008

PIER

View full text Add to dashboard Cite

Abstract-To improve the parallel efficiency in the case of the finegrained FDTD computing on PC cluster, the concept of "two level parallelization on PC cluster" is presented, and a high performance MPI-OpenMP hybrid FDTD algorithm is developed. In the hybrid algorithm, MPI is used in conjunction with OpenMP multithreading to achieve two level parallelism of the data and tasks at the basis of the domain decomposition FDTD method. Besides, to enhance the flexibility of the parallel FDTD, the interpolation between subspaces is also discussed. The simulation example of a printed antenna for automobile is given. Computations are performed for different numbers of PCs and contrasted with two conventional parallel FDTD algorithms on PC cluster. The results show that with the decrease of the computational granularity on each computer, the novel algorithm is more efficient, and moreover, it can also lessen the influence of the sub-domains virtual topology on the parallel FDTD performance.

show abstract

Performance comparison of pure MPI vs hybrid MPI-OpenMP parallelization models on SMP clusters

Cited by 43 publications

References 14 publications

Invasive Compute Balancing for Applications with Shared and Hybrid Parallelization

Invasive Compute Balancing for Applications with Shared and Hybrid Parallelization

Speeding Up Distributed MapReduce Applications Using Hardware Accelerators

A Novel FDTD Approach Featuring Two-Level Parallelization on Pc Cluster

Contact Info

Product

Resources

About