Implementing a parallel C++ runtime system for scalable parallel systems

Malony, Allen D.; Mohr, Bernd; Beckman, Peter H.; Gannon, Dennis; Yang, Shiming; Bodin, François; Kesavan, Suraj P.

doi:10.1145/169627.169802

Cited by 27 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The work by Michael L. Nelson et al [7] is further enhanced by Gregory V. Wilson et al [8] by adopting the built-in library methods in C++ and also by E. Arjomandi et al [9] using the default inheritance strategies of modern object-oriented programming languages. Considering only the OO development strategies for making the code parallelization a good number of research attempts were made as A. Krishnamurthy et al [10] using the split-C method, P. A. Buhr et al [11] and Xining Li et al [12] using the default concurrency control by the OO programming languages, S. Shelly et al [13] using the inter-process communication between objects, F. Bodin et al [14] by deploying customization to the runtime. Also, many of the parallel research outcomes have demonstrated the use of a newer programming language to take the maximum advantage of the GPU as shown by Matthew Fluet et al [15].…”

Section: Outcomes From the Parallel Researchesmentioning

confidence: 99%

Convert Component Based Serial Code to Parallel to an Automatic Framework

2024

IJFMR

View full text Add to dashboard Cite

— With the increase of modular software development to reduce the delivery time, the increase of parallel processing is also increasing. The serial software codes, in the form of component or without the components, cannot leverage the performance of the modern hardware capabilities. Hence majority of the recent software developments are happening considering the parallel processing of the source code. The parallel processing of the source code is not only the responsibility of hardware or GPUs, rather the source codes are also to be written in specific form to take the benefits of parallel processors. Subsequently, not constrained to the improvement in the calculation structure, the utilization of parallel execution of the projects is likewise to be considered. GPUs are ordinarily utilized handling units to accelerate the application execution in diversion improvement. The GPUs can be used to parallelize the application execution to achieve the clock used to the most extreme. The real test is to plan or re-structure the application code from customary sequential programming dialects to the parallel codes, which can take the benefits of GPU centres. Nevertheless, developing the parallel software code components demand a higher development skill from the developer’s community, which is difficult to obtain. Also, apart from the newer applications, there are many legacy software components which are also under the demand for conversion in the parallel form. A good number of parallel research attempts are carried out to convert the serial code into a parallel form code, nonetheless, the attempts were manual and needs a very high amount of programming and the application APIs knowledge. Hence, this research attempts to build a framework to automatically convert the serial source code components into parallel source code components to be plugged in to the parallel applications without misplacing the benefits of component-based and the benefits from parallel processing of software codes. The proposed novel framework demonstrates a nearly 97% reduction of code execution time.

show abstract

Section: Outcomes From the Parallel Researchesmentioning

confidence: 99%

Convert Component Based Serial Code to Parallel to an Automatic Framework

2024

IJFMR

View full text Add to dashboard Cite

show abstract

“…A number of such languages were developed in the late 1980s and early 1990s, including Fortran D [356,475], Vienna Fortran [188,1018], CM Fortran [921], C* [432], data-parallel C, and PC++ [641]. These research efforts were the precursors of informal standardization activities leading to High Performance Fortran (HPF) [468].…”

Section: Data-parallel Programming In High Performance Fortranmentioning

confidence: 99%

Sourcebook of parallel computing

Dongarra¹,

Foster

Fox

et al. 2003

Choice Reviews Online

View full text Add to dashboard Cite

“…Roughly speaking, there are two major directions in those efforts. In the Fortran world, HPF 11 21, Fortran D 11 31, Vienna Fortran [6] and others are developed, in the C world, HPC 1211, ICC++ [7], MPC++ [14], pC++ [16], EC++ [20] and others are in progress.…”

Section: Comparisons and Conclusionmentioning

confidence: 99%

High-level data parallel programming in Promoter

Besch¹,

Bi²,

Enskonatus³

et al.

Proceedings Second International Workshop on High-Level Parallel Programming Models and Supportive Environments

View full text Add to dashboard Cite

Implementing realistic scientific applications on parallel platforms requires a high-level, problem-adequate and flexible programming environment. The hybrid system PRO-MOTER pursues a two-level approach allowing easy and flexible programming at both language and library levels. The core concept of PROMOTER'S language model is its highly abstract and unijied concept of data and communication structures. The paper briefly addresses the programming model, but focuses on implementation aspects of the compiler and runtime system. Finally, pe$ormance results are given, evaluating the eficiency of the PROMOTER system.

show abstract

Implementing a parallel C++ runtime system for scalable parallel systems

Cited by 27 publications

References 6 publications

Convert Component Based Serial Code to Parallel to an Automatic Framework

Convert Component Based Serial Code to Parallel to an Automatic Framework

Sourcebook of parallel computing

High-level data parallel programming in Promoter

Contact Info

Product

Resources

About