In high-performance computing, storage is a shared resource and used by all users with many different application requirements and knowledge of storage. Consequently, the optimal storage configuration varies according...
详细信息
ISBN:
(纸本)9781728109121
In high-performance computing, storage is a shared resource and used by all users with many different application requirements and knowledge of storage. Consequently, the optimal storage configuration varies according to the I/O behavior of each application. While system logs are helpful resources in understanding the storage behavior, it is nontrivial for each user to analyze the logs and adjust complex configurations. Even for experienced users, it is difficult to understand the full stack of I/O systems and find the optimal configuration for the specific application. In this work, we analyzed the I/O activities of CORI which is an HPC system in National Energy Research Scientific computing Center (NERSC). The result of our analysis shows that most users do not adjust storage configurations and use the default settings. Also, it shows that only a few applications are executed repeatedly in the HPC environment. Based on this result, we have developed DCA-I/O, a dynamic distributed file system configuration adjustment algorithm, which utilizes system log information and widely adapted rules to adjust storage configurations automatically without any user intervention. DCA-I/) utilizes existing system logs and does not require any modifications in code or an additional library. To demonstrate the effectiveness of DCA-I/O, we have performed experiments using I/O kernels of the real applications in both isolated small-sized Lustre environment and CORI. Our experimental result shows that the use of our scheme can lead to improvements in the performance of INC applications by up to 75% in an isolated environment and 50% in a real HPC environment without user intervention.
The proceedings contain 92 papers. The topics discussed include: parallel domain decomposition methods for ray-tracing on multi-cores and multi-processors;the research of Q learning-based estimation of distribution al...
ISBN:
(纸本)9780769544151
The proceedings contain 92 papers. The topics discussed include: parallel domain decomposition methods for ray-tracing on multi-cores and multi-processors;the research of Q learning-based estimation of distribution algorithm;chaotic relaxation asynchronous iterative parallel algorithm and its implementation;coupling the parareal algorithm with the waveform relaxation method for the solution of differential algebraic equations;the research of feature selection of text classification based on integrated learning algorithm;formal support for cyber physical system specification using aspect-oriented approach;moment problem of G-frames in Hilbert spaces;a highly practicable distributed architecture of network stream media;parallel implementation of fractal image compression in web service environment;a new carry-free adder model for ternary optical computer;and a web page classification algorithm based on link information.
Reuse-based software engineering is becoming the main approach for developing business and commercial systems. Service choreographies support the reuse-based service-oriented philosophy in that they represent a powerf...
详细信息
ISBN:
(纸本)9781450359337
Reuse-based software engineering is becoming the main approach for developing business and commercial systems. Service choreographies support the reuse-based service-oriented philosophy in that they represent a powerful and flexible approach to realize systems by (possibly) reusing services and composing them in a fully distributed way. A key enabler for the realization of choreographies is the ability to (i) reuse existing applications and services that can play the role of choreography participants, (ii) aid developers in writing the code of those participants whose roles cannot be covered through reuse, and (iii) automatically synthesize the coordination logic required for their correct interaction. The contribution of this paper is the definition and realization of a novel approach to the synthesis of service choreographies that allows developers to just fill-in-the-blank of automatically generated code templates of single (uncovered) choreography tasks, without the need of considering all the message flows specified by the choreography and the related distributed coordination issues.
In this paper the defined contribution pension is assumed to be invested in three assets: a risk-free, a low-risk and the other high-risk assets. The key problem is how to allocate the assets dynamically to outperform...
详细信息
In this paper the defined contribution pension is assumed to be invested in three assets: a risk-free, a low-risk and the other high-risk assets. The key problem is how to allocate the assets dynamically to outperform the market. Due to stochastic process of the prices of risky assets, the problem is essentially a stochastic dynamic programming which is described by a Hamilton-Jacobi-Bellman (HJB) equation. The difference scheme and the strategy iteration are used to numerically solve the HJB equation. Finally the weights of the optimal portfolio are given and analyzed.
By use of statistical theory, an unknown large biological model is considered, the habits of unknown large organisms are studied, the effects of unknown large organisms on ecological environment are evaluated, the ene...
详细信息
By use of statistical theory, an unknown large biological model is considered, the habits of unknown large organisms are studied, the effects of unknown large organisms on ecological environment are evaluated, the energy and heat consumption required by themselves are calculated, and the ambient area needed to be carried out with *** analysis and calculation results in this paper make people better and faster to understand unknown large organisms.
Fibrous porous media plays an important role in aerosol or small particle filtration. The filtration of fibrous porous media is a composite process, of which the filtration performance is affected by many parameters a...
详细信息
Fibrous porous media plays an important role in aerosol or small particle filtration. The filtration of fibrous porous media is a composite process, of which the filtration performance is affected by many parameters and is quite difficult to be predicted precisely. In this paper the parameters affecting the filtration performance are re-examined. Especially, the influences of fiber sizes and the porosity of the porous media to the filtration efficiency are analyzed further.
The increase in scale provided by distributedcomputing systems has expanded scientific discovery and engineering solutions. Stochastic modeling with Performance Evaluation Process Algebra (PEPA) has been used to eval...
详细信息
ISBN:
(数字)9781728189468
ISBN:
(纸本)9781728189475
The increase in scale provided by distributedcomputing systems has expanded scientific discovery and engineering solutions. Stochastic modeling with Performance Evaluation Process Algebra (PEPA) has been used to evaluate the robustness of static resource allocations in parallel and distributedcomputing systems. These evaluations have previously been performed through the PEPA Plug-In for the Eclipse Integrated Development Environment and have been limited by factors that include: i) the size and complexity of the underlying, in-use PEPA model, ii) a small number of resource allocation models available for analysis, and iii) the human interaction necessary to configure the PEPA Eclipse Plug-In, thus limiting potential automation. As the size and complexity of the underlying PEPA models increases, the number of states to be evaluated for each model also greatly increases, leading to a case of state space explosion. In this work, we validate the Imperial PEPA Compiler (IPC) as a replacement for the PEPA Eclipse Plug-In for the robustness analysis of resource allocations. We make available an implementation of the IPC as a Singularity container, as part of a larger online repository of PEPA resources. We then develop and test a programmatic method for generating PEPA models for resource allocations. When combined with our IPC container, this method allows automated analysis of resource allocation models at scale. The use of the IPC allows the evaluation of larger models than it is possible when using the PEPA Eclipse Plug-In. Moreover, the increases in scale in both model size and number of models, support the development of improved makespan targets for robustness metrics, including those among applications subject to perturbations at runtime, as found in typical parallel and distributedcomputing environments.
Proper planning of the commercial vehicle supply chain is conducive to improving work efficiency and reducing transportation costs. In this paper, the Dijkstra algorithm is used to solve the neighboring weight matrix ...
详细信息
Proper planning of the commercial vehicle supply chain is conducive to improving work efficiency and reducing transportation costs. In this paper, the Dijkstra algorithm is used to solve the neighboring weight matrix between different cities. The 0-1 planning model is established by distance, fixed cost and demand, and the optimal total cost scheme is calculated by genetic algorithm. The simulated annealing algorithm is used to solve the problem of minimum road damage in the case of a certain increase in total cost, which is convenient for the transportation company to make rational adjustments of the commercial vehicle supply chain network.
Algebraic connectivity and Fiedler vector are extremely important global state parameters of the network. Algebraic connectivity determines the convergence speed of distributed algorithms in the network, and Fiedler v...
详细信息
Algebraic connectivity and Fiedler vector are extremely important global state parameters of the network. Algebraic connectivity determines the convergence speed of distributed algorithms in the network, and Fiedler vector determines the importance of each node in the network. At present, distributed algorithms for estimating algebraic connectivity are limited to fixed network systems, and the convergence of estimation algorithms and optimization of convergence are not studied in depth. Aiming at the above problems, this paper applies the algebraic connectivity estimation algorithm to the random opportunistic network, and then analyzes the convergence of the algorithm and deduces the convergence speed formula of the algorithm, and finally optimizes the convergence of the algorithm.
暂无评论