An approach for non-intrusively adding malleable fork/join parallelism into ordinary JavaBean compliant applications

Mateos, Cristian; Zunino, Alejandro; Campo, Marcelo

doi:10.1016/j.cl.2009.12.003

Cited by 10 publications

(18 citation statements)

References 65 publications

(92 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are several general-purpose frameworks that have been used to process large-scale graphs in link prediction. Frameworks such as MapReduce [6], Fork-Join [15] or RDD (Spark) [16] have been applied to various graph-related processing problems [17][18][19]. Other frameworks, like Pregel [8] or GraphLab [20] are specifically designed for graph-based algorithms [21,22].…”

Section: Related Workmentioning

confidence: 99%

“…For example, path-based link prediction algorithms fit naturally under the divide-and-conquer strategy and, thus, a Fork-Join [15] framework (also known as Split and Merge or Split and Reduce in some frameworks 1 ) may be a good fit. The main entity in a distributed FJ algorithm is the FJ job, which represents the computation as a whole.…”

Section: Distributed Partitioned Mergementioning

confidence: 99%

See 1 more Smart Citation

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

Corbellini

Godoy

Mateos

et al. 2018

Future Generation Computer Systems

Self Cite

View full text Add to dashboard Cite

Large-scale graphs have become ubiquitous in social media. Computer-based recommendations in these huge graphs pose challenges in terms of algorithm design and resource usage efficiency when processing recommendations in distributed computing environments. Moreover, recommendation algorithms for graphs, particularly link prediction algorithms, have different requirements depending of the way the underlying graph is traversed. Path-based algorithms usually perform traversals in different directions to build a large ranking of vertices to recommend, whereas random walk-based algorithms build an initial subgraph and perform several iterations on those vertices to compute the final ranking. In this work, we propose a distributed graph processing framework called Distributed Partitioned Merge (DPM), which supports both types of algorithms and we compare its performance and resource usage w.r.t. two relevant frameworks, namely Fork-Join and Pregel. In our experiments, we show that in most tests DPM outperforms both Pregel and Fork-Join in terms of recommendation time, with a minor penalization in network usage in some scenarios. *Revised Manuscript with source files (Word document) Click here to download Revised Manuscript with source files (Word document): fgcs.pdf Click here to view linked Referenc

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Distributed Partitioned Mergementioning

confidence: 99%

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

Corbellini

Godoy

Mateos

et al. 2018

Future Generation Computer Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…Important results regarding parallelization methods and development of techniques to facilitate software mobility have emerged in the last decade . These advances are aimed to traditional parallel and distributed infrastructures, which rely on fixed servers connected for predictable and stable intervals of time.…”

Section: Related Workmentioning

confidence: 99%

“…After selecting the operating system, we analyzed Java and Android native code as the programming languages to do the experiments. We decided to use Java because of its increasing popularity and its interesting features; that is, it has been recognized as being a useful language to develop and execute resource‐intensive code by researchers . We are interested in this feature because one of our goals is to contribute to execute scientific applications in mobile devices.…”

Section: Introductionmentioning

confidence: 99%

Improving scientific application execution on android mobile devices via code refactorings

2016

Self Cite

View full text Add to dashboard Cite

The increasing number of mobile devices with ever-growing capabilities makes them useful for running scientific applications. However, these applications have high computational demands, whereas mobile devices have limited capabilities when compared with non-mobile devices. More importantly, mobile devices rely on batteries for their power supply. We initially measure the battery consumption of different versions of known micro-benchmarks representing common programming primitives found in scientific applications. Then, we analyze the performance of such micro-benchmarks in CPU-intensive mobile applications. We apply good programming practices and code refactorings to reduce battery consumption of scientific mobile applications. Our results show the reduction in energy usage from applying these refactorings to three scientific applications, and we consequently propose guidelines for high-performance computing applications. Our focus is on Android, the dominant mobile operating system. As a long-term contribution, our results represent one more step in the progress towards hybrid distributed infrastructures comprising fixed and mobile nodes, that is, the so-called mobile grids.In general, this device is more powerful than a smartphone. Furthermore, smartphones and tablets offer interesting hardware features such as sensors, GPS, and accelerometers.Additionally, it is important to mention the accelerated evolution of mobile devices. We can observe the constant progress in smartphones by comparing the Samsung Galaxy S, which was launched on September 9, 2010, with its successor the Samsung Galaxy SII. The Samsung Galaxy S used the Samsung S5PC110 processor, which combined a 45-nm 1-GHz ARM Cortex-A8-based CPU with 512 MB of RAM. Also, it has a 2 GB of internal storage. Regarding the battery life, it has a talk time of up to 6 h and a standby time of up to 12.5 days. The Samsung Galaxy S supports Bluetooth 3.0, WiFi 802.11b/g/n, and 3G data up to 7.2 Mbit/s. In contrast, the Galaxy S II has a 1.2-GHz dual-core processor, 1 GB of RAM, and 16 GB of internal mass storage. It has a battery standby time of up to 10.5 days and talk time of up to 8 h. Subsequent models that followed have even more capacity.Battery consumption in mobile devices is one of the most important problems in mobile computing research. The example earlier shows that while computational capacity has increased rapidly, battery power has increased slowly [3,4]. This paper aims to assess the energy capabilities of mobile devices using different micro-benchmarks commonly found in scientific applications and to show the impact of several known code refactorings to save battery in mobile devices. This means that we chose refactorings that have been proposed to improve maintainability, readability, and so on and we evaluated their impact in battery consumption. Currently mobile devices are not only devices for connection/access to external computing resources, but they are becoming core nodes of computational infrastructures in scientific projects [1]. Ther...

show abstract

“…It was proposed by Lea [25] in 2000, and has been widely applied in scientific fields [26,27]. It was integrated, as a standard parallel framework, into Java7 [28] and released in July 2011.…”

Section: Fork/join Frameworkmentioning

confidence: 99%

A modular parallelization framework for power flow transfer analysis of large-scale power systems

Cheng

Luo

Shen

et al. 2017

J. Mod. Power Syst. Clean Energy

View full text Add to dashboard Cite

Power flow transfer (PFT) analysis under various anticipated faults in advance is important for securing power system operations. In China, PSD-BPA software is the most widely used tool for power system analysis, but its input/output interface is easily adapted for PFT analysis, which is also difficult due to its computationally intensity. To solve this issue, and achieve a fast and accurate PFT analysis, a modular parallelization framework is developed in this paper. Two major contributions are included. One is several integrated PFT analysis modules, including parameter initialization, fault setting, network integrity detection, reasonableness identification and result analysis. The other is a parallelization technique for enhancing computation efficiency using a Fork/Join framework. The proposed framework has been tested and validated by the IEEE 39 bus reference power system. Furthermore, it has been applied to a practical power network with 11052 buses and 12487 branches in the Yunnan Power Grid of China, providing decision support for large-scale power system analysis.

show abstract

An approach for non-intrusively adding malleable fork/join parallelism into ordinary JavaBean compliant applications

Cited by 10 publications

References 65 publications

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

Improving scientific application execution on android mobile devices via code refactorings

A modular parallelization framework for power flow transfer analysis of large-scale power systems

Contact Info

Product

Resources

About