Pub Date : 2019-07-01DOI: 10.4018/IJGHPC.2019070102
Khyati Ahlawat, A. Chug, A. Singh
Imbalanced datasets are the ones with uneven distribution of classes that deteriorates classifier's performance. In this paper, SVM classifier is combined with K-Means clustering approach and a hybrid approach, Hy_SVM_KM is introduced. The performance of proposed method is also empirically evaluated using Accuracy and FN Rate measure and compared with existing methods like SMOTE. The results have shown that the proposed hybrid technique has outperformed traditional machine learning classifier SVM in mostly datasets and have performed better than known pre-processing technique SMOTE for all datasets. The goal of this article is to extend capabilities of popular machine learning algorithms and adapt it to meet the challenges of imbalanced big data classification. This article can provide a baseline study for future research on imbalanced big datasets classification and provides an efficient mechanism to deal with imbalanced nature big dataset with modified SVM classifier and improves the overall performance of the model.
{"title":"Empirical Evaluation of Map Reduce Based Hybrid Approach for Problem of Imbalanced Classification in Big Data","authors":"Khyati Ahlawat, A. Chug, A. Singh","doi":"10.4018/IJGHPC.2019070102","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019070102","url":null,"abstract":"Imbalanced datasets are the ones with uneven distribution of classes that deteriorates classifier's performance. In this paper, SVM classifier is combined with K-Means clustering approach and a hybrid approach, Hy_SVM_KM is introduced. The performance of proposed method is also empirically evaluated using Accuracy and FN Rate measure and compared with existing methods like SMOTE. The results have shown that the proposed hybrid technique has outperformed traditional machine learning classifier SVM in mostly datasets and have performed better than known pre-processing technique SMOTE for all datasets. The goal of this article is to extend capabilities of popular machine learning algorithms and adapt it to meet the challenges of imbalanced big data classification. This article can provide a baseline study for future research on imbalanced big datasets classification and provides an efficient mechanism to deal with imbalanced nature big dataset with modified SVM classifier and improves the overall performance of the model.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"17 1","pages":"23-45"},"PeriodicalIF":1.0,"publicationDate":"2019-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75350387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.4018/IJGHPC.2019040103
A. Savyanavar, V. Ghorpade
A mobile grid (MG) consists of interconnected mobile devices which are used for high performance computing. Fault tolerance is an important property of mobile computational grid systems for achieving superior arrangement reliability and faster recovery from failures. Since the failure of the resources affects task execution fatally, fault tolerance service is essential to achieve QoS requirement in MG. The faults which occur in MG are link failure, node failure, task failure, limited bandwidth etc. Detecting these failures can help in better utilisation of the resources and timely notification to the user in a MG environment. These failures result in loss of computational results and data. Many algorithms or techniques were proposed for failure handling in traditional grids. The authors propose a checkpointing based failure handling technique which will improve arrangement reliability and failure recovery time for the MG network. Experimentation was conducted by creating a grid of ubiquitously available Android-based mobile phones.
{"title":"Application Checkpointing Technique for Self-Healing From Failures in Mobile Grid Computing","authors":"A. Savyanavar, V. Ghorpade","doi":"10.4018/IJGHPC.2019040103","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019040103","url":null,"abstract":"A mobile grid (MG) consists of interconnected mobile devices which are used for high performance computing. Fault tolerance is an important property of mobile computational grid systems for achieving superior arrangement reliability and faster recovery from failures. Since the failure of the resources affects task execution fatally, fault tolerance service is essential to achieve QoS requirement in MG. The faults which occur in MG are link failure, node failure, task failure, limited bandwidth etc. Detecting these failures can help in better utilisation of the resources and timely notification to the user in a MG environment. These failures result in loss of computational results and data. Many algorithms or techniques were proposed for failure handling in traditional grids. The authors propose a checkpointing based failure handling technique which will improve arrangement reliability and failure recovery time for the MG network. Experimentation was conducted by creating a grid of ubiquitously available Android-based mobile phones.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"242 1","pages":"50-62"},"PeriodicalIF":1.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76978473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.4018/IJGHPC.2019040105
Ramasubbareddy Somula, Eduardo Oliveros, T. Cucinotta, S. Phillips, Xiaoyu Yang, Jia Chen, Cui-xia Ma, Hongan Wang, Hai-yan Yang
Day to day the usage of mobile devices (MD) is growing in people's lives. But still the MD is limited in terms of memory, battery life time, processing capacity. In order to overcome these issues, the new emerging technology named mobile cloud computing (MCC) has been introduced. The offloading mechanism execute the resource intensive application on the remote cloud to save both the battery utilization and execution time. But still the high latency challenges in MCC need to be addressed by executing resource intensive task at nearby resource cloud server. The key challenge is to find optimal cloudlet to execute task to save computation time. In this article, the authors propose a Round Robin algorithm based on cloudlet selection in heterogeneous MCC system. This article considers both load and distance of server to find optimal cloudlet and minimize waiting time of the user request at server queue. Additionally, the authors provide mathematical evaluation of the algorithm and compare with existing load balancing algorithms.
{"title":"A Load and Distance Aware Cloudlet Selection Strategy in Multi-Cloudlet Environment","authors":"Ramasubbareddy Somula, Eduardo Oliveros, T. Cucinotta, S. Phillips, Xiaoyu Yang, Jia Chen, Cui-xia Ma, Hongan Wang, Hai-yan Yang","doi":"10.4018/IJGHPC.2019040105","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019040105","url":null,"abstract":"Day to day the usage of mobile devices (MD) is growing in people's lives. But still the MD is limited in terms of memory, battery life time, processing capacity. In order to overcome these issues, the new emerging technology named mobile cloud computing (MCC) has been introduced. The offloading mechanism execute the resource intensive application on the remote cloud to save both the battery utilization and execution time. But still the high latency challenges in MCC need to be addressed by executing resource intensive task at nearby resource cloud server. The key challenge is to find optimal cloudlet to execute task to save computation time. In this article, the authors propose a Round Robin algorithm based on cloudlet selection in heterogeneous MCC system. This article considers both load and distance of server to find optimal cloudlet and minimize waiting time of the user request at server queue. Additionally, the authors provide mathematical evaluation of the algorithm and compare with existing load balancing algorithms.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"18 1","pages":"85-102"},"PeriodicalIF":1.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74089689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.4018/IJGHPC.2019040101
Lei Li, Yilin Wang, Lianwen Jin, Xin Zhang, Huiping Qin
Workload prediction is important for automatic scaling of resource management, and a high accuracy of workload prediction can reduce the cost and improve the resource utilization in the cloud. But, the task request is usually random mutation, so it is difficult to achieve more accurate prediction result for single models. Thus, to improve the prediction result, the authors proposed a novel two-stage workload prediction model based on artificial neural networks (ANNs), which is composed of one classification model and two prediction models. On the basis of the first-order gradient feature, the model can categorize the workload into two classes adaptively. Then, it can predict the workload by using the corresponding prediction neural network models according to the classification results. The experiment results demonstrate that the suggested model can achieve more accurate workload prediction compared with other models.
{"title":"Two-Stage Adaptive Classification Cloud Workload Prediction Based on Neural Networks","authors":"Lei Li, Yilin Wang, Lianwen Jin, Xin Zhang, Huiping Qin","doi":"10.4018/IJGHPC.2019040101","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019040101","url":null,"abstract":"Workload prediction is important for automatic scaling of resource management, and a high accuracy of workload prediction can reduce the cost and improve the resource utilization in the cloud. But, the task request is usually random mutation, so it is difficult to achieve more accurate prediction result for single models. Thus, to improve the prediction result, the authors proposed a novel two-stage workload prediction model based on artificial neural networks (ANNs), which is composed of one classification model and two prediction models. On the basis of the first-order gradient feature, the model can categorize the workload into two classes adaptively. Then, it can predict the workload by using the corresponding prediction neural network models according to the classification results. The experiment results demonstrate that the suggested model can achieve more accurate workload prediction compared with other models.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"1 1","pages":"1-23"},"PeriodicalIF":1.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90775384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.4018/IJGHPC.2019040102
Kirit J. Modi, Sanjay Garg, S. Chaudhary
RESTful web services have evolved based on REST architectural design and gained popularity because of their inherent simplicity and suitability features in comparison with SOAP-based web services. Moreover, linked open data (LOD) provides a uniform data model for RESTful web services which in turn avoids manual intervention of users to perform tasks such as, searching, selection, and integration. Researchers have worked on LOD based RESTful web services searching, selection and composition but focused on individual basis though they are interrelated tasks. This article presents an integrated framework and approach to automate the discovery, selection and composition of RESTful Web services using linked open data to provide an efficient composition solution. We work with RDF descriptions to express the state of linked data resources on which SPARQL queries would be applied for the extraction, filtering and integration of RESTful services. Use case scenarios of population information systems and healthcare recommendation systems are presented as a proof of concept with necessary results.
{"title":"An Integrated Framework for RESTful Web Services Using Linked Open Data","authors":"Kirit J. Modi, Sanjay Garg, S. Chaudhary","doi":"10.4018/IJGHPC.2019040102","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019040102","url":null,"abstract":"RESTful web services have evolved based on REST architectural design and gained popularity because of their inherent simplicity and suitability features in comparison with SOAP-based web services. Moreover, linked open data (LOD) provides a uniform data model for RESTful web services which in turn avoids manual intervention of users to perform tasks such as, searching, selection, and integration. Researchers have worked on LOD based RESTful web services searching, selection and composition but focused on individual basis though they are interrelated tasks. This article presents an integrated framework and approach to automate the discovery, selection and composition of RESTful Web services using linked open data to provide an efficient composition solution. We work with RDF descriptions to express the state of linked data resources on which SPARQL queries would be applied for the extraction, filtering and integration of RESTful services. Use case scenarios of population information systems and healthcare recommendation systems are presented as a proof of concept with necessary results.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"15 1","pages":"24-49"},"PeriodicalIF":1.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73697617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-04-01DOI: 10.4018/IJGHPC.2019040104
M. Shelar, S. Sane, V. Kharat
Server virtualization is a well-known technique for virtual machine (VM) placement and consolidation and has been studied extensively by several researchers. This article presents a novel approach called aiCloud that advocates segmentation of hosts or physical machines (PMs) into four different classes that facilitates quick selection of PMs to reduce the time required to search host machines, called host search time (HST). The framework also introduces VM_Acceptance_State, a condition that avoids host overloading, which leads to significant reduction of SLA time per active host (SLATAH) that in turn reduces SLA violation (SLAV). The performance of aiCloud has been compared with other approaches using standard workload traces. Empirical evaluation presented shows that aiCloud has least HST and outperforms other approaches in terms of SLA violations and ESV (Energy and SLA Violation) and therefore may be an attractive strategy for efficient management of cloud resources.
{"title":"A Novel Energy Efficient and SLA-Aware Approach for Cloud Resource Management","authors":"M. Shelar, S. Sane, V. Kharat","doi":"10.4018/IJGHPC.2019040104","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019040104","url":null,"abstract":"Server virtualization is a well-known technique for virtual machine (VM) placement and consolidation and has been studied extensively by several researchers. This article presents a novel approach called aiCloud that advocates segmentation of hosts or physical machines (PMs) into four different classes that facilitates quick selection of PMs to reduce the time required to search host machines, called host search time (HST). The framework also introduces VM_Acceptance_State, a condition that avoids host overloading, which leads to significant reduction of SLA time per active host (SLATAH) that in turn reduces SLA violation (SLAV). The performance of aiCloud has been compared with other approaches using standard workload traces. Empirical evaluation presented shows that aiCloud has least HST and outperforms other approaches in terms of SLA violations and ESV (Energy and SLA Violation) and therefore may be an attractive strategy for efficient management of cloud resources.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"35 1","pages":"63-84"},"PeriodicalIF":1.0,"publicationDate":"2019-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87310975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.4018/IJGHPC.2019010105
Ihtisham Ali, S. Bagchi
Workflow is an essential mechanism for the automation of processes in distributed transactional systems, including mobile distributed systems. The workflow modeling enables the composition of process activities along with respective conditions, data flow and control flow dependencies. The workflow partitioning methods are used to create sub-partitions by grouping processes on the basis of activities, data flow and control flow dependencies. Mobile distributed systems consisting of heterogeneous computing devices require optimal workflow decomposition. In general, the workflow partitioning is a NP-complete problem. This article presents a comparative study and detailed analysis of workflow decomposition techniques based on graphs, petri nets and topological methods. A complete taxonomy of the basic decomposition techniques is presented. A detailed qualitative and quantitative analysis of these decomposition techniques are explained. The comparative analysis presented in this article provides an insight to inherent algorithmic complexities of respective decomposition approaches. The qualitative parametric analysis would help in determining the suitability of workflow applicability in different computing environments involving static and dynamic nodes. Furthermore, the authors have presented a novel framework for workflow decomposition based on multiple parametric parameters for mobile distributed systems.
{"title":"A Comparative Study and Algorithmic Analysis of Workflow Decomposition in Distributed Systems","authors":"Ihtisham Ali, S. Bagchi","doi":"10.4018/IJGHPC.2019010105","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019010105","url":null,"abstract":"Workflow is an essential mechanism for the automation of processes in distributed transactional systems, including mobile distributed systems. The workflow modeling enables the composition of process activities along with respective conditions, data flow and control flow dependencies. The workflow partitioning methods are used to create sub-partitions by grouping processes on the basis of activities, data flow and control flow dependencies. Mobile distributed systems consisting of heterogeneous computing devices require optimal workflow decomposition. In general, the workflow partitioning is a NP-complete problem. This article presents a comparative study and detailed analysis of workflow decomposition techniques based on graphs, petri nets and topological methods. A complete taxonomy of the basic decomposition techniques is presented. A detailed qualitative and quantitative analysis of these decomposition techniques are explained. The comparative analysis presented in this article provides an insight to inherent algorithmic complexities of respective decomposition approaches. The qualitative parametric analysis would help in determining the suitability of workflow applicability in different computing environments involving static and dynamic nodes. Furthermore, the authors have presented a novel framework for workflow decomposition based on multiple parametric parameters for mobile distributed systems.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"59 1","pages":"71-100"},"PeriodicalIF":1.0,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84501100","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.4018/IJGHPC.2019010101
Ilangovan Sangaiya, A. V. A. Kumar
In data mining, people require feature selection to select relevant features and to remove unimportant irrelevant features from a original data set based on some evolution criteria. Filter and wrapper are the two methods used but here the authors have proposed a hybrid feature selection method to take advantage of both methods. The proposed method uses symmetrical uncertainty and genetic algorithms for selecting the optimal feature subset. This has been done so as to improve processing time by reducing the dimension of the data set without compromising the classification accuracy. This proposed hybrid algorithm is much faster and scales well to the data set in terms of selected features, classification accuracy and running time than most existing algorithms.
{"title":"A Hybrid Feature Selection Method for Effective Data Classification in Data Mining Applications","authors":"Ilangovan Sangaiya, A. V. A. Kumar","doi":"10.4018/IJGHPC.2019010101","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019010101","url":null,"abstract":"In data mining, people require feature selection to select relevant features and to remove unimportant irrelevant features from a original data set based on some evolution criteria. Filter and wrapper are the two methods used but here the authors have proposed a hybrid feature selection method to take advantage of both methods. The proposed method uses symmetrical uncertainty and genetic algorithms for selecting the optimal feature subset. This has been done so as to improve processing time by reducing the dimension of the data set without compromising the classification accuracy. This proposed hybrid algorithm is much faster and scales well to the data set in terms of selected features, classification accuracy and running time than most existing algorithms.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"7 1","pages":"1-16"},"PeriodicalIF":1.0,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86407916","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.4018/IJGHPC.2019010102
Huming Zhu, Peidao Li, P. Zhang, Zheng Luo
A ranking support vector machine (RSVM) is a typical pairwise method of learning to rank, which is effective in ranking problems. However, the training speed of RSVMs are not satisfactory, especially when solving large-scale data ranking problems. Recent years, many-core processing units (graphics processing unit (GPU), Many Integrated Core (MIC)) and multi-core processing units have exhibited huge superiority in the parallel computing domain. With the support of hardware, parallel programming develops rapidly. Open Computing Language (OpenCL) and Open Multi-Processing (OpenMP) are two of popular parallel programming interfaces. The authors present two high-performance parallel implementations of RSVM, an OpenCL version implemented on multi-core and many-core platforms, and an OpenMP version implemented on multi-core platform. The experimental results show that the OpenCL version parallel RSVM achieved considerable speedup on Intel MIC 7110P, NVIDIA Tesla K20M and Intel Xeon E5-2692v2, and it also shows good portability.
排序支持向量机(RSVM)是一种典型的两两学习排序方法，在排序问题中非常有效。然而，rsvm的训练速度并不令人满意，特别是在解决大规模数据排序问题时。近年来，多核处理单元(图形处理单元(GPU)、多集成核(MIC))和多核处理单元在并行计算领域显示出巨大的优势。在硬件的支持下，并行编程得到了迅速的发展。开放计算语言(OpenCL)和开放多处理(OpenMP)是两种流行的并行编程接口。作者提出了两种RSVM的高性能并行实现，一种是在多核和多核平台上实现的OpenCL版本，一种是在多核平台上实现的OpenMP版本。实验结果表明，OpenCL版本并行RSVM在Intel MIC 7110P、NVIDIA Tesla K20M和Intel Xeon E5-2692v2上取得了相当大的加速，并表现出良好的可移植性。
{"title":"A High Performance Parallel Ranking SVM with OpenCL on Multi-core and Many-core Platforms","authors":"Huming Zhu, Peidao Li, P. Zhang, Zheng Luo","doi":"10.4018/IJGHPC.2019010102","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019010102","url":null,"abstract":"A ranking support vector machine (RSVM) is a typical pairwise method of learning to rank, which is effective in ranking problems. However, the training speed of RSVMs are not satisfactory, especially when solving large-scale data ranking problems. Recent years, many-core processing units (graphics processing unit (GPU), Many Integrated Core (MIC)) and multi-core processing units have exhibited huge superiority in the parallel computing domain. With the support of hardware, parallel programming develops rapidly. Open Computing Language (OpenCL) and Open Multi-Processing (OpenMP) are two of popular parallel programming interfaces. The authors present two high-performance parallel implementations of RSVM, an OpenCL version implemented on multi-core and many-core platforms, and an OpenMP version implemented on multi-core platform. The experimental results show that the OpenCL version parallel RSVM achieved considerable speedup on Intel MIC 7110P, NVIDIA Tesla K20M and Intel Xeon E5-2692v2, and it also shows good portability.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"30 1","pages":"17-28"},"PeriodicalIF":1.0,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"79355947","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-01-01DOI: 10.4018/IJGHPC.2019010104
Mohsin Altaf Wani, Manzoor Ahmad
Modern GPUs perform computation at a very high rate when compared to CPUs; as a result, they are increasingly used for general purpose parallel computation. Determining if a statically optimal binary search tree is an optimization problem to find the optimal arrangement of nodes in a binary search tree so that average search time is minimized. Knuth's modification to the dynamic programming algorithm improves the time complexity to O(n2). We develop a multiple GPU-based implementation of this algorithm using different approaches. Using suitable GPU implementation for a given workload provides a speedup of up to four times over other GPU based implementations. We are able to achieve a speedup factor of 409 on older GTX 570 and a speedup factor of 745 is achieved on a more modern GTX 1060 when compared to a conventional single threaded CPU based implementation.
{"title":"Statically Optimal Binary Search Tree Computation Using Non-Serial Polyadic Dynamic Programming on GPU's","authors":"Mohsin Altaf Wani, Manzoor Ahmad","doi":"10.4018/IJGHPC.2019010104","DOIUrl":"https://doi.org/10.4018/IJGHPC.2019010104","url":null,"abstract":"Modern GPUs perform computation at a very high rate when compared to CPUs; as a result, they are increasingly used for general purpose parallel computation. Determining if a statically optimal binary search tree is an optimization problem to find the optimal arrangement of nodes in a binary search tree so that average search time is minimized. Knuth's modification to the dynamic programming algorithm improves the time complexity to O(n2). We develop a multiple GPU-based implementation of this algorithm using different approaches. Using suitable GPU implementation for a given workload provides a speedup of up to four times over other GPU based implementations. We are able to achieve a speedup factor of 409 on older GTX 570 and a speedup factor of 745 is achieved on a more modern GTX 1060 when compared to a conventional single threaded CPU based implementation.","PeriodicalId":43565,"journal":{"name":"International Journal of Grid and High Performance Computing","volume":"32 1","pages":"49-70"},"PeriodicalIF":1.0,"publicationDate":"2019-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76428020","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：