MPI Scaling Up for Powerlist Based Parallel Programs

Niculescu, Virginia; Bufnea, Darius; Sterca, Adrian

doi:10.1109/empdp.2019.8671597

Cited by 5 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The JPLF framework offers the possibility to develop efficient implementations based on some templates that facilitate a simple and robust implementation. Using it, we proved that this can lead to efficient implementations on shared and distributed memory platforms of many divide-and-conquer algorithms [45,[48][49][50]. The templates based on PowerList allow decompositions of degree two (decomposition into two subproblems), while the variants based on PList allow decomposition with different arities [51].…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

On Generalizing Divide and Conquer Parallel Programming Pattern

Niculescu¹

2022

Mathematics

Self Cite

View full text Add to dashboard Cite

(1) Background: Structuring is important in parallel programming in order to master its complexity, and this structuring could be achieved through programming patterns and skeletons. Divide-and-conquer computation is essentially defined by a recurrence relation that links the solution of a problem to the solutions of subproblems of the same type, but of smaller sizes. This pattern allows the specification of different types of computations, and so it is important to provide a general specification that comprises all its cases. We intend to prove that the divide-and-conquer pattern could be generalized such that to comprise many of the other parallel programming patterns, and in order to prove this, we provide a general formulation of it. (2) Methods: Starting from the proposed generalized specification of the divide-and-conquer pattern, the computation of the pattern is analyzed based on its stages: decomposition, base-case and composition. Examples are provided, and different execution models are analyzed. (3) Results: a general functional specification is provided for a divide-and-conquer pattern and based on it, and we prove that this general formulation could be specialized through parameters’ instantiating into other classical parallel programming patterns. Based on the specific stages of the divide-and-conquer, three classes of computations are emphasized. In this context, an equivalent efficient bottom-up computation is formally proved. Associated models of executions are emphasized and analyzed based on the three classes of divide-and-conquer computations. (4) Conclusion: A more general definition of the divide-and-conquer pattern is provided, and this includes an arity list for different decomposition degrees, a level of recursion, and also an alternative solution for the cases that are not trivial but allow other approaches (sequential or parallel) that could lead to better performance. Together with the associated analysis of patterns equivalence and optimized execution models, this provides a general formulation that is useful both at the semantic level and implementation level.

show abstract

Section: Discussionmentioning

confidence: 99%

“…When distributed memory is used, data communication adds an important overhead, that should be optimized through data packaging, data serialization, or by combing reading with data distribution, or/and data aggregation with data writing. Examples of frameworks that treat these problems are those reported in [44,45].…”

Section: Synchronization and Data Managementmentioning

confidence: 99%

On Generalizing Divide and Conquer Parallel Programming Pattern

Niculescu¹

2022

Mathematics

Self Cite

View full text Add to dashboard Cite

show abstract

“…The ability to use multiple cluster nodes for execution could be attained by introducing MPI based execution of the functions [27]. This assures the needed scalability for a framework that works with regular data sets of very large sizes.…”

Section: Mpi Executionmentioning

confidence: 99%

“…A general architecture is proposed for this kind of frameworks -MEDUGA (Model-Executors-DataManager-UserInteracter-GranularityBalancer-metricsAnalyser). This design is exemplified on a concrete framework -JPLF: Java Parallel Lists Framework [29,27]. This paper is an extension of the conference paper [30], in which the patterns and software development principles used for the JPLF implementation were analysed.…”

Section: Introductionmentioning

confidence: 99%

Reflections on the Design of Parallel Programming Frameworks

Niculescu

Sterca

Loulergue

2021

Communications in Computer and Information Science

Self Cite

View full text Add to dashboard Cite

Parallel programming is much more complex and difficult than sequential programming, and it is therefore more challenging to achieve the same software quality in a parallel context. High-level parallel programming models, if implemented as software frameworks, could increase productivity and reliability. Important requirements such as extensibility and adaptability for different platforms are required for such a framework, and this paper reflects on these requirements and their relation to the software engineering methodologies that could put them in practice. All these are exemplified on a Java framework -JPLF; this is a high-level parallel programming approach being based on the model brought by the PowerLists associated theories, and it respects the analysed requirements. The design of JPLF is analysed by explaining the design choices and highlighting the design patterns and design principles applied.

show abstract