Pub Date : 2015-07-20DOI: 10.1109/HPCSim.2015.7237047
Omar Khedher, M. Jarraya
Cloud Computing has empowered the academic communities by reducing the IT infrastructure administration and cost. As one of several cloud computing solutions, Virtual Cloud Laboratory (VCL) has improved many universities experience by providing hand-on labs for each course per class. VCL has multiple deployment requirements. Although it represents a simplified architecture design, treating a large size VCL deployment might be challenging. In addition, putting the performance of resource allocation policies and different scheduling algorithms under scope could help to quantify the workload running in a real and expanded VCL environment. We aim through this study at identifying the limitations of the default scheduler in VCL and propose an enhanced scheduler. By simulating a VCL cloud computing environment, it might be possible to prove different scheduling algorithms and their impact from performance perspective.
{"title":"Improved scheduling algorithm in VCL Cloud computing environment on CloudSim","authors":"Omar Khedher, M. Jarraya","doi":"10.1109/HPCSim.2015.7237047","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237047","url":null,"abstract":"Cloud Computing has empowered the academic communities by reducing the IT infrastructure administration and cost. As one of several cloud computing solutions, Virtual Cloud Laboratory (VCL) has improved many universities experience by providing hand-on labs for each course per class. VCL has multiple deployment requirements. Although it represents a simplified architecture design, treating a large size VCL deployment might be challenging. In addition, putting the performance of resource allocation policies and different scheduling algorithms under scope could help to quantify the workload running in a real and expanded VCL environment. We aim through this study at identifying the limitations of the default scheduler in VCL and propose an enhanced scheduler. By simulating a VCL cloud computing environment, it might be possible to prove different scheduling algorithms and their impact from performance perspective.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"177 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116639670","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237092
J. Raik
A dedicated fault monitoring network supplemented with fault management capabilities is required for controlling the dynamic reconfiguration and resource isolation within many-core systems as well as for keeping track of the system health status and reporting it to the software layer. It is imperative to develop such cross-layer fault management capabilities to be deployed in safety-critical domains but also in applications where access for maintenance would be time-consuming and costly, e.g. base stations in the telecommunications area. European universities (Tallinn University of Technology, Graz University of Technology and the University of Twente) have teamed up with IBM, the German Space Centre DLR and two SMEs (Recore Systems and Testonica Lab) in a new EU funded research project IMMORTAL (Integrated Modelling, Fault Management, Verification and Reliable Design Environment for Cyber-Physical Systems) in a quest to improve reliability and extend the lifetime of emerging many-core based cyber-physical systems.
{"title":"Immortalizing many-core systems early experiences of the horizon 2020 action IMMORTAL","authors":"J. Raik","doi":"10.1109/HPCSim.2015.7237092","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237092","url":null,"abstract":"A dedicated fault monitoring network supplemented with fault management capabilities is required for controlling the dynamic reconfiguration and resource isolation within many-core systems as well as for keeping track of the system health status and reporting it to the software layer. It is imperative to develop such cross-layer fault management capabilities to be deployed in safety-critical domains but also in applications where access for maintenance would be time-consuming and costly, e.g. base stations in the telecommunications area. European universities (Tallinn University of Technology, Graz University of Technology and the University of Twente) have teamed up with IBM, the German Space Centre DLR and two SMEs (Recore Systems and Testonica Lab) in a new EU funded research project IMMORTAL (Integrated Modelling, Fault Management, Verification and Reliable Design Environment for Cyber-Physical Systems) in a quest to improve reliability and extend the lifetime of emerging many-core based cyber-physical systems.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125884137","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237099
Pragya Jain, A. Goel, S. Gupta
For the purposes of operating and managing complex cloud storage infrastructure, there is a need of monitoring it effectively and efficiently. Data is required to be collected from major components and analyzed for improving system performance and resource utilization. Swift is an open source object storage cloud system by OpenStack, also popularly used by many key commercial cloud providers. In this paper, we present a detailed list of attributes required for monitoring Swift storage infrastructure. We see that some of the processes associated with object storage need to be monitored and also resources the storage software utilizes for its working. We note that infrastructure associated with storage and its usage is required to be monitored constantly for providing better services. Accordingly, we categorize infrastructure monitoring into components and list parameters associated with them. Our monitoring list is beneficial for system administrator when selecting monitoring software for their system. During development of any new monitoring software, proposed monitoring list aids developer in selecting functionality in contrast to specifying fresh requirements.
{"title":"Requirement checklist for infrastructure monitoring of swift","authors":"Pragya Jain, A. Goel, S. Gupta","doi":"10.1109/HPCSim.2015.7237099","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237099","url":null,"abstract":"For the purposes of operating and managing complex cloud storage infrastructure, there is a need of monitoring it effectively and efficiently. Data is required to be collected from major components and analyzed for improving system performance and resource utilization. Swift is an open source object storage cloud system by OpenStack, also popularly used by many key commercial cloud providers. In this paper, we present a detailed list of attributes required for monitoring Swift storage infrastructure. We see that some of the processes associated with object storage need to be monitored and also resources the storage software utilizes for its working. We note that infrastructure associated with storage and its usage is required to be monitored constantly for providing better services. Accordingly, we categorize infrastructure monitoring into components and list parameters associated with them. Our monitoring list is beneficial for system administrator when selecting monitoring software for their system. During development of any new monitoring software, proposed monitoring list aids developer in selecting functionality in contrast to specifying fresh requirements.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126062454","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237048
M. Schiller, A. Knoll, M. Mocker, T. Eibert
Due to the complexity of Vehicular Ad Hoc Networks, future driving assistance systems need to be validated through virtual test drives in a simulated environment. An accurate modeling of the vehicle-to-vehicle communication channel is crucial to enable a precise evaluation of such network-aware applications. Since existing ray-based methods cause long computation times, a new parallel GPU-based ray-launching simulation method is presented. The algorithmic improvements allow a high utilization of the GPU computing power, which results in significantly faster simulations while achieving high accuracy. The validation of the simulation results against real-world measurements showed a high level of agreement.
{"title":"GPU accelerated ray launching for high-fidelity virtual test drives of VANET applications","authors":"M. Schiller, A. Knoll, M. Mocker, T. Eibert","doi":"10.1109/HPCSim.2015.7237048","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237048","url":null,"abstract":"Due to the complexity of Vehicular Ad Hoc Networks, future driving assistance systems need to be validated through virtual test drives in a simulated environment. An accurate modeling of the vehicle-to-vehicle communication channel is crucial to enable a precise evaluation of such network-aware applications. Since existing ray-based methods cause long computation times, a new parallel GPU-based ray-launching simulation method is presented. The algorithmic improvements allow a high utilization of the GPU computing power, which results in significantly faster simulations while achieving high accuracy. The validation of the simulation results against real-world measurements showed a high level of agreement.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115577121","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237067
N. Melab, Rudi Leroy, M. Mezmaz, D. Tuyttens
Many combinatorial optimization problems are modeled in practice as permutation-based ones. We have recently proposed a new data structure called IVM dedicated to those problems. IVM is memory efficient in terms of size and management time for solving large permutation problems using Branch-and-Bound (B&B) algorithm. We believe that those memory properties make IVM well-suited for Many Integrated Cores (MIC) architecture. This paper deals with the parallel design and implementation of the B&B algorithm on MIC architectures using private IVM-based work stealing. The proposed approach has been extensively experimented on an Intel Xeon Phi 5110P using several instances of the Flow-Shop scheduling permutation problem. The reported results show that the IVM-based work stealing approach is about 10 times faster than the linked-list traditionally used for parallel B&B.
{"title":"Parallel Branch-and-Bound using private IVM-based work stealing on Xeon Phi MIC coprocessor","authors":"N. Melab, Rudi Leroy, M. Mezmaz, D. Tuyttens","doi":"10.1109/HPCSim.2015.7237067","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237067","url":null,"abstract":"Many combinatorial optimization problems are modeled in practice as permutation-based ones. We have recently proposed a new data structure called IVM dedicated to those problems. IVM is memory efficient in terms of size and management time for solving large permutation problems using Branch-and-Bound (B&B) algorithm. We believe that those memory properties make IVM well-suited for Many Integrated Cores (MIC) architecture. This paper deals with the parallel design and implementation of the B&B algorithm on MIC architectures using private IVM-based work stealing. The proposed approach has been extensively experimented on an Intel Xeon Phi 5110P using several instances of the Flow-Shop scheduling permutation problem. The reported results show that the IVM-based work stealing approach is about 10 times faster than the linked-list traditionally used for parallel B&B.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122749683","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237023
H. Shoukourian, T. Wilde, A. Auweter, A. Bode
Efficient scheduling is crucial for time and cost-effective utilization of compute resources especially for high end systems. A variety of factors need to be considered during the scheduling decisions. Power variation across the compute resources of homogeneous large-scale systems has not been considered so far. This paper discusses the impact of the power variation for parallel application scheduling. It addresses the problem of finding the optimal resource configuration for a given application that will minimize the amount of consumed energy, under pre-defined constraints on application execution time and instantaneous average power consumption. This paper presents an efficient algorithm to do so, which also considers the existing power diversity among the compute nodes (modified also at different operating CPU frequencies) of a given homogeneous High Performance Computing system. Based on this algorithm, the paper presents a plug-in, referred to as Configuration Adviser, which operates on top of a given resource management and scheduling system to advise on energy-wise optimal resource configuration for a given application, execution using which, will adhere to the specified execution time and power consumption constraints. The main goal of this plug-in is to enhance the current resource management and scheduling tools for the support of power capping for future Exascale systems, where a data center might not be able to provide cooling or electrical power for system peak consumption but only for the expected power bands.
{"title":"Power variation aware Configuration Adviser for scalable HPC schedulers","authors":"H. Shoukourian, T. Wilde, A. Auweter, A. Bode","doi":"10.1109/HPCSim.2015.7237023","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237023","url":null,"abstract":"Efficient scheduling is crucial for time and cost-effective utilization of compute resources especially for high end systems. A variety of factors need to be considered during the scheduling decisions. Power variation across the compute resources of homogeneous large-scale systems has not been considered so far. This paper discusses the impact of the power variation for parallel application scheduling. It addresses the problem of finding the optimal resource configuration for a given application that will minimize the amount of consumed energy, under pre-defined constraints on application execution time and instantaneous average power consumption. This paper presents an efficient algorithm to do so, which also considers the existing power diversity among the compute nodes (modified also at different operating CPU frequencies) of a given homogeneous High Performance Computing system. Based on this algorithm, the paper presents a plug-in, referred to as Configuration Adviser, which operates on top of a given resource management and scheduling system to advise on energy-wise optimal resource configuration for a given application, execution using which, will adhere to the specified execution time and power consumption constraints. The main goal of this plug-in is to enhance the current resource management and scheduling tools for the support of power capping for future Exascale systems, where a data center might not be able to provide cooling or electrical power for system peak consumption but only for the expected power bands.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"89 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116612474","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237103
Artur Yuzhanin, I. Gankevich, E. Stepanov, V. Korkhov
In this paper the Monte Carlo methods of the Asian option pricing are considered. Among them are pricing method with path integral and partial differential equation. Simulation algorithms running on the CPU sequentially and algorithms running on the GPU in parallel using the CUDA technology were analyzed and compared.
{"title":"Efficient Asian option pricing with CUDA","authors":"Artur Yuzhanin, I. Gankevich, E. Stepanov, V. Korkhov","doi":"10.1109/HPCSim.2015.7237103","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237103","url":null,"abstract":"In this paper the Monte Carlo methods of the Asian option pricing are considered. Among them are pricing method with path integral and partial differential equation. Simulation algorithms running on the CPU sequentially and algorithms running on the GPU in parallel using the CUDA technology were analyzed and compared.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129298653","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237111
R. Meijer, R. Cushing, C. D. Laat, P. Jackson, S. Klous, R. Koning, M. Makkes, A. Meerwijk
In the invited talk “Car2x with SDN, NFV and supercomputers” we report about how our past work with SDN [1, 2] allows the design of a smart mobility fieldlab in the huge parking lot the Amsterdam Arena. We explain how we can engineer and test software that handle the complex conditions of the Car2X case. The talk starts by describing the engineering challenges that developers of smart car telecommunications and computing infrastructures face. We concentrate on the development of software defined networks (SDN) that support smart cars optimally and securely over a heterogeneous, dynamic and developing ICT infrastructure. The goal here is to enable smart cars to profit maximally from any bit of information available from fixed and moving objects as well as persons. For example, in a low tech situation, the lack of other options makes that one has to fall back to GPRS to download only traffic jam locations. As a contrast, in an advanced telecom environment video streams from multiple cars are transmitted via 5G pico cells to computers a few millisecond nearby. These computers fuse the video information to generate a local traffic model. In an ultimate situation, cars use all the communications infrastructures that are available including the numerous WiFi hotspots, all generations of mobile telecommunications, the developing car-to-car communications technologies and even the smart phone of a passing person. SDN technologies deal with the complexities of such communication environment.
{"title":"Car2x with software defined networks, network functions virtualization and supercomputers technical and scientific preparations for the Amsterdam Arena telecoms fieldlab","authors":"R. Meijer, R. Cushing, C. D. Laat, P. Jackson, S. Klous, R. Koning, M. Makkes, A. Meerwijk","doi":"10.1109/HPCSim.2015.7237111","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237111","url":null,"abstract":"In the invited talk “Car2x with SDN, NFV and supercomputers” we report about how our past work with SDN [1, 2] allows the design of a smart mobility fieldlab in the huge parking lot the Amsterdam Arena. We explain how we can engineer and test software that handle the complex conditions of the Car2X case. The talk starts by describing the engineering challenges that developers of smart car telecommunications and computing infrastructures face. We concentrate on the development of software defined networks (SDN) that support smart cars optimally and securely over a heterogeneous, dynamic and developing ICT infrastructure. The goal here is to enable smart cars to profit maximally from any bit of information available from fixed and moving objects as well as persons. For example, in a low tech situation, the lack of other options makes that one has to fall back to GPRS to download only traffic jam locations. As a contrast, in an advanced telecom environment video streams from multiple cars are transmitted via 5G pico cells to computers a few millisecond nearby. These computers fuse the video information to generate a local traffic model. In an ultimate situation, cars use all the communications infrastructures that are available including the numerous WiFi hotspots, all generations of mobile telecommunications, the developing car-to-car communications technologies and even the smart phone of a passing person. SDN technologies deal with the complexities of such communication environment.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121185235","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237093
M. Moghaddam, A. Yamamoto, Cristinel Ababei
We investigate dynamic voltage and frequency scaling (DVFS) as a mechanism for dynamic reliability management (DRM) of chip multiprocessors (CMPs). The proposed DRM scheme operates as a control technique whose objective is to drive the operation of the CMP such that reliability changes towards a desired target. While the chip multiprocessor is continuously monitored and reliability is estimated in real time, the voltage and frequency of different cores in the CMP are dynamically adjusted such that reliability converges towards the target. When the temperature of cores increases and thus reliability degrades, the proposed DRM scheme throttles selectively the frequency of the cores with the highest temperature. This is turn, leads to a lower power dissipation in those cores whose temperature decreases, thereby improving reliability. We leverage existing simulation and estimation tools to develop the proposed DRM scheme. Simulations results show that the proposed DRM scheme provides an effective way to tradeoff reliability and performance.
{"title":"Investigation of DVFS based dynamic reliability management for chip multiprocessors","authors":"M. Moghaddam, A. Yamamoto, Cristinel Ababei","doi":"10.1109/HPCSim.2015.7237093","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237093","url":null,"abstract":"We investigate dynamic voltage and frequency scaling (DVFS) as a mechanism for dynamic reliability management (DRM) of chip multiprocessors (CMPs). The proposed DRM scheme operates as a control technique whose objective is to drive the operation of the CMP such that reliability changes towards a desired target. While the chip multiprocessor is continuously monitored and reliability is estimated in real time, the voltage and frequency of different cores in the CMP are dynamically adjusted such that reliability converges towards the target. When the temperature of cores increases and thus reliability degrades, the proposed DRM scheme throttles selectively the frequency of the cores with the highest temperature. This is turn, leads to a lower power dissipation in those cores whose temperature decreases, thereby improving reliability. We leverage existing simulation and estimation tools to develop the proposed DRM scheme. Simulations results show that the proposed DRM scheme provides an effective way to tradeoff reliability and performance.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114629715","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 : 2015-07-20DOI: 10.1109/HPCSim.2015.7237071
J. Nonaka, K. Ono, M. Fujita
Binary-Swap is a parallel image compositing algorithm based on recursive vector halving and distance doubling, and works efficiently when the number of processes is exactly a power-of-two (2n). Several power-of-two converting approaches for Binary-Swap have been proposed. Among them, the Telescope method, based on the Binary Blocks algorithm, has been shown as the most promising approach. The Telescope method decomposes an entire set of processes into blocks of power-of-two size and merges the smaller blocks into larger blocks in stepwise fashion. This block merging process corresponds to the communication and computational overhead of the conversion, and since it can only merge one block per stage, it becomes inefficient as the number of binary blocks increases. In this paper, we focus on a single-stage conversion method using the 3-2 and 2-1 elimination approaches. The original scheduling method, proposed by Rabenseifner et al., is limited to an odd number of processes since it always schedules a single 3-2 elimination per conversion. Taking into consideration that the 3-2 elimination can be optimized on modern HPC systems, which can overlap the communication and computation, we propose 234 Scheduling for scheduling multiple 3-2 eliminations per conversion. The multiple 3-2 elimination scheduling enlarges the application range by enabling its use on an even number of processes. We evaluated 234 Scheduling applied to Binary-Swap on the K computer, which is a modern parallel HPC system, and confirmed its effectiveness.
{"title":"234 scheduling of 3-2 and 2-1 eliminations for parallel image compositing using non-power-of-two number of processes","authors":"J. Nonaka, K. Ono, M. Fujita","doi":"10.1109/HPCSim.2015.7237071","DOIUrl":"https://doi.org/10.1109/HPCSim.2015.7237071","url":null,"abstract":"Binary-Swap is a parallel image compositing algorithm based on recursive vector halving and distance doubling, and works efficiently when the number of processes is exactly a power-of-two (2n). Several power-of-two converting approaches for Binary-Swap have been proposed. Among them, the Telescope method, based on the Binary Blocks algorithm, has been shown as the most promising approach. The Telescope method decomposes an entire set of processes into blocks of power-of-two size and merges the smaller blocks into larger blocks in stepwise fashion. This block merging process corresponds to the communication and computational overhead of the conversion, and since it can only merge one block per stage, it becomes inefficient as the number of binary blocks increases. In this paper, we focus on a single-stage conversion method using the 3-2 and 2-1 elimination approaches. The original scheduling method, proposed by Rabenseifner et al., is limited to an odd number of processes since it always schedules a single 3-2 elimination per conversion. Taking into consideration that the 3-2 elimination can be optimized on modern HPC systems, which can overlap the communication and computation, we propose 234 Scheduling for scheduling multiple 3-2 eliminations per conversion. The multiple 3-2 elimination scheduling enlarges the application range by enabling its use on an even number of processes. We evaluated 234 Scheduling applied to Binary-Swap on the K computer, which is a modern parallel HPC system, and confirmed its effectiveness.","PeriodicalId":134009,"journal":{"name":"2015 International Conference on High Performance Computing & Simulation (HPCS)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127676274","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：