2017 International Conference on High Performance Computing & Simulation (HPCS)最新文献

{"title":"Reducing Randomization in the Power of Two Choices Load Balancing Algorithm","authors":"Félix García Carballeira, A. Calderón","doi":"10.1109/HPCS.2017.62","DOIUrl":"https://doi.org/10.1109/HPCS.2017.62","url":null,"abstract":"This paper proposes a new version of the Power of Two Choices, SQ(d), load balancing algorithm that improves the performance of the classical model based on the power of two choices randomized load balancing. This model considers jobs that arrive to a dispatcher as a Poisson stream of rate λn, λ 1, at a set of n servers. Using the power of two choices, the dispatcher chooses for each job some d constant independently and uniformly from the n servers in a random way, and sends the job to the server with the fewest number of jobs. This algorithm offers advantage over the load balancing based on shortest queue discipline, because it offers a good performance, and reduces the overhead over the servers and over the communication network. In this paper, we propose a new version, Shortest Queue of d with Randomization and Round Robin Policies, SQ-RR(d), that combines randomization techniques and static local balancing based on round robin policy. In this new version the dispatcher chooses the d servers as follows: one is selected using round robin policy and the d - 1 servers are chosen independently and uniformly in a random way from the η servers. Then, the dispatcher sends the job to the server with the fewest number of jobs. We demonstrate with an analytical approximation of this approach, that this new version improves the performance obtained with the classical solution for d 2, and obtains similar results for d 2, included systems at 99 percent of capacity. Furthermore, we provide simulations that demonstrate the analytical approximation developed and show the behavior of this algorithm with realistic workloads based on Google datacenter traces.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"146 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126896596","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}

{"title":"Performance Analysis with Cache-Aware Roofline Model in Intel Advisor","authors":"Diogo Marques, Helder Duarte, A. Ilic, L. Sousa, Roman Belenov, P. Thierry, Zakhar A. Matveev","doi":"10.1109/HPCS.2017.150","DOIUrl":"https://doi.org/10.1109/HPCS.2017.150","url":null,"abstract":"The recent increase in the complexity of processor architectures imposes significant challenges when designing and optimizing the execution of real-world applications, even on general-purpose hardware. To help in this process, tools for fast and insightful visualization of architecture and application execution bottlenecks are particularly useful for computer architects and application engineers, such as the recently proposed Cache-aware Roofline Model (CARM). CARM represents an insightful architecture performance model that provides a simple and intuitive way of visually representing the limits of parallel processing on contemporary multi-core processors with complex memory hierarchy. In its recent updates, Intel Advisor integrated performance CARM into its workflow. Intel Advisor is a powerful tool that helps application developers to extract the full potential performance out of a processor architecture, by analyzing applications and providing hints on parallelization, vectorization and memory access improvements. Therefore, when coupled with CARM, Intel Advisor Roofline represents a complete analysis and visualization framework for application characterization, optimization, and development. This paper focuses on introducing the CARM analysis methodology within Intel Advisor, by also showcasing the usability of other Advisor features. For this purpose, a set of 10 applications from different benchmark suits were analyzed on the state-of-the-art hardware platform in order to uncover the most critical bottlenecks and possible optimization steps to overcome them. By following the optimization guidelines given by Intel Advisor Roofline, the performance of several application kernels was improved for up to 6.43 times when compared to the unoptimized versions.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"254 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117319207","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}

{"title":"A Parallel Library for Social Media Analytics","authors":"Loris Belcastro, F. Marozzo, D. Talia, Paolo Trunfio","doi":"10.1109/HPCS.2017.105","DOIUrl":"https://doi.org/10.1109/HPCS.2017.105","url":null,"abstract":"Social media analysis is a fast growing research area aimed at extracting useful information from huge amounts of data generated by social media users. This work presents a Java library, called ParSoDA (Parallel Social Data Analytics), which can be used for developing parallel data analysis applications based on the extraction of useful knowledge from large dataset gathered from social networks. The library aims at reducing the programming skills necessary to implement scalable social data analysis applications. To reach this goal, ParSoDA defines a general structure for a social data analysis application that includes a number of configurable steps, and provides a predefined (but extensible) set of functions that can be used for each step. The paper describes the ParSoDA library and presents two case studies to assess its usability and scalability.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"134 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122435104","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}

{"title":"A Parallel I/O Behavior Model for HPC Applications Using Serial I/O Libraries","authors":"P. Gomez-Sanchez, Sandra Méndez, Dolores Rexachs, E. Luque","doi":"10.1109/HPCS.2017.45","DOIUrl":"https://doi.org/10.1109/HPCS.2017.45","url":null,"abstract":"Analyzing and understanding an application's Input/Output (I/O) access patterns provides key information to gain insight into how the behavior of an application affects its performance in different systems. In this paper, we propose a portable model to represent the I/O behavior of parallel applications that use serial I/O libraries, as part of a more holistic model for I/O of parallel applications. The model allows actions such as replication of the application behavior on different HPC systems and evaluation of the I/O performance without running the real application. In this paper, we evaluate the portability of the proposed model for MADbench2 and ABYSS- P in four HPC systems. We analyze the impact of the parallel system configuration in a Cloud environment for the ABYSS-P application by using the proposed I/O behavior model.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114388616","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}

{"title":"A Directive-Based Approach to Perform Persistent Checkpoint/Restart","authors":"Marcos Maroñas, Sergi Mateo, Vicencc Beltran, E. Ayguadé","doi":"10.1109/HPCS.2017.72","DOIUrl":"https://doi.org/10.1109/HPCS.2017.72","url":null,"abstract":"Exascale platforms require support for resilience capabilities due to increasing numbers of components and associated error rates. In this paper, we present a new directive-based approach to perform application-level checkpoint/restart in a simplified and portable way. We propose a solution based on compiler directives, similar to OpenMP, that allows users to easily specify the state of the application that has to be saved and restored. This leaves the tedious and error-prone serialization and deserialization activities to our library, which relies on SCR/FTI to perform scalable and efficient I/O operations. Our results, based on several benchmarks and two large applications, reveal no additional overhead compared to the direct use of FTI and SCR checkpoint/restart libraries. Apart from that, our portable approach significantly increases the programmability reducing the number of code lines required to perform checkpoint/restart in an average of ≈ 82% and ≈ 94%, for FTI and SCR respectively.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133194050","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}

{"title":"Collaborative Information Retrieval Model Based on Fuzzy Clustering","authors":"F. Naouar, L. Hlaoua, Mohamed Nazih Omri","doi":"10.1109/HPCS.2017.80","DOIUrl":"https://doi.org/10.1109/HPCS.2017.80","url":null,"abstract":"The collaborative approach has shown interest in several fields of application, particularly in information retrieval to satisfy a need for shared information. Despite this collaboration, the search for relevant information is always a tedious task as long as the mass of information continues to increase, part of which is a source, while other parties represent comments on these sources. It is obvious that nowadays we witness an explosion of multimedia documents so that multimedia information retrieval techniques remain insufficient to satisfy the needs of the user despite the collaborative framework: multimedia-type documents cannot be rich in information and more specifically the video documents. We consider, therefore, annotations as a new source of information. In addition to their relevance, we notice that annotations express generally brief ideas using some words that they cannot be comprehensible independently of his context. To use them, a classification is considered necessary. The emergence of new annotations should be considered and therefore the classification should be extended. A centroid is determined in a virtual way to represent each annotation class. From where, the interest to use the fuzzy classification to know which elements can belong to several clusters. It consists, in a calculation of the center of gravity of all the existing classes. This is the reason why; we proposed a fuzzy clustering-based annotation. In the experiments, we tried to consider a relevance feedback system based on confidence network considering new relevant classified annotations as a source of information. To validate this model, we have carried out a set of experiments and we have obtained encouraging results.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"180 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131650269","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}

{"title":"Improving the Network Performance of a Container-Based Cloud Environment for Hadoop Systems","authors":"C. Rista, Dalvan Griebler, Carlos A. F. Maron, L. G. Fernandes","doi":"10.1109/HPCS.2017.97","DOIUrl":"https://doi.org/10.1109/HPCS.2017.97","url":null,"abstract":"Cloud computing has emerged as an important paradigm to improve resource utilization, efficiency, flexibility, and the pay-per-use billing structure. However, cloud platforms cause performance degradations due to their virtualization layer and may not be appropriate for the requirements of high-performance applications, such as big data. This paper tackles the problem of improving network performance in container-based cloud instances to create a viable alternative to run network intensive Hadoop applications. Our approach consists of deploying link aggregation via the IEEE 802.3ad standard to increase the available bandwidth and using LXC (Linux Container) cloud instances to create a Hadoop cluster. In order to evaluate the efficiency of our approach and the overhead added by the container-based cloud environment, we ran a set of experiments to measure throughput, latency, bandwidth utilization, and completion times. The results prove that our approach adds minimal overhead in cloud environment as well as increases throughput and reduces latency. Moreover, our approach demonstrates a suitable alternative for running Hadoop applications, reducing completion times up to 33.73%.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134407895","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}

{"title":"Evaluation of SDN-based Conflict Avoidance between Data Staging and Inter-Process Communication","authors":"Arata Endo, Ryoichi Jingai, S. Date, Y. Kido, S. Shimojo","doi":"10.1109/HPCS.2017.48","DOIUrl":"https://doi.org/10.1109/HPCS.2017.48","url":null,"abstract":"High-performance computing systems are currently used by many scientists and researchers with Job Management Systems (JMS) that are deployed over high-performance computing systems. To achieve high performance in high-performance computing, systems communication performance must be improved in the interconnect. JMS sometimes adopts the staging, witch is a method to transfer data used by an application program to a local high-speed storage of computing nodes so that the access time to the data is reduced. In this type of high-performance computing systems, the inter-process and the staging communication traffic coexist in an interconnect. In this paper, we investigate the mutual influence between the inter-process and the staging communication traffic and explore a way to exclude the influence of the staging and inter-process communications by targeting a cluster system with a fat-tree interconnect. For this purpose, we propose two types of conflict avoidance methods between the staging communication traffic and the inter-process communication traffic; namely, a link separation conflict avoidance method and a link sharing conflict avoidance method. The link separation conflict avoidance method separates interconnect links into the staging communication traffic and the inter-process communication traffic when the staging execution takes place. The latter proposed method allocates a route to each traffic on the shared interconnect while taking the bandwidth usage into account. Both proposed methods use Software Defined Networking to get the bandwidth usage and to control both types of traffic. We performed an experimental simulation on our deployed virtual cluster system built on a single machine to investigate how each proposed method behaves when the staging communication and the inter-process communication take place in a fat-tree interconnect. In the simulation, the staging time, application execution time and job execution time were measured for each proposed method and for the Equal-Cost Multi-Path method as the existing load distributed method was applied. The result shows that both of the proposed methods improve the performance of the staging and the inter-process communication.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134106282","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}

{"title":"A Methodology for Soft Errors Detection and Automatic Recovery","authors":"Jorge Villamayor, Dolores Rexachs, E. Luque, Diego Montezanti, A. D. Giusti, M. Naiouf","doi":"10.1109/HPCS.2017.71","DOIUrl":"https://doi.org/10.1109/HPCS.2017.71","url":null,"abstract":"Handling faults is a growing concern in HPC; higher error rates, larger detection intervals and silent faults are expected in the future. It is projected that, in exascale systems, errors will occur several times a day, and they will propagate to generate errors that will range from process crashes to corrupted results because of undetected errors. In this article, we propose a methodology that improves system reliability against transient faults, when running parallel message-passing applications. The proposed solution, based on process replication, has the goal of helping programmers and users of parallel scientific applications to achieve reliable executions with correct results. This work presents a characterization of the strategy, defining its behavior in the presence of faults and modeling the temporal costs of employing it. As a result, we show its efficacy and viability to tolerate transient faults in HPC systems.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"150 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123739197","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}

{"title":"FLAME GPU: Complex System Simulation Framework","authors":"P. Richmond, M. Chimeh","doi":"10.1109/HPCS.2017.12","DOIUrl":"https://doi.org/10.1109/HPCS.2017.12","url":null,"abstract":"FLAME GPU is an agent based simulation frame-work that utilises the parallel architecture of Graphic Processing Unit (GPU) to enable real time model interaction and visualisation. In this paper, we provide an overview of the features of FLAME GPU and demonstrate its efficiency as a parallel agent based simulation platform. FLAME GPU abstracts the complexity of the GPU architecture from the users by offering a high level modelling syntax based on a formal state machine representation. A flocking model is presented showing how a simple multi-agent system is modelled.","PeriodicalId":115758,"journal":{"name":"2017 International Conference on High Performance Computing & Simulation (HPCS)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128300773","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}