ISSN 0236-235X (P)
ISSN 2311-2735 (E)

Bookmark

Next issue

3
Publication date:
16 September 2019
-->

Articles of journal № 4 at 2012 year.

Order result by:
Public date | Title | Authors

1. Choosing computational system for scientific problems [№4 за 2012 год]
Author: Шабанов Б.М.
Visitors: 8702
The article describes mapping of computer architecture to application programs. This problem is important when in is needed to choose computer system for certain programs. Formalization of choice of computer system for solving scientific and engineering problems is discussed. Study of efficiency of the program on a cluster with multicore processors and GPUs is presented. It is considered a synchronous model of the program with exchanges: between cores, between the processor and the accelerator and between the computational nodes, time estimations for data transfer time are provided. The following methods of determining parameters of numerical model are analyzed: profiling programs on a computer system, simulation of the program on the system, evaluation model of the program and the system. Some typical cases of data exchange are considered: exchange with the «neighbors» (e.g., exchange between the nodes of a multidimensional mesh) and collective communications (one to all, all to one). To provide data for solving problems of this kind in the JSCC RAS there was developed a set of benchmarks representing different areas of science, also there was developed test program that measures floating point performance of cores, memory performance, file operations performance and communications performance. Three problems of computer system choice are considered – determination of the system components system for solving certain problems to reach: maximum performance, minimum cost of the system or maximum performance at a fixed price. Specific features of costs minimization are discussed. The described approach was used in the selection of architectures for JSCC RAS high-performance systems such as MVS-10BM, MVS-6000IM, MVS-100K

2. Identity federation of the research and educational networks [№4 за 2012 год]
Authors: Овсянников А.П., Савин Г.И., Шабанов Б.М.
Visitors: 7419
The article is devoted to identity federations of the research and educational networks. Research and educational networks contain a lot of informational and computing resources. These resources are in a dif-ferent organizational, administrative and public supervision. Scientific resources require authentication, for example, to monitor compliance with intellectual property rights, the definition of user rights, etc. Authorization methods with user authentication by his institution are commonly used. Identity federation is a group of institutions had an agreements on mutual access to their resources, developed a common policy and user identity framework, and implemented them on the organizational and technical level. Identity federation have operational team, steering group and user support service. Due to diversity and international level of the research identity federations cooperates to interfederation. The article describes eduroam and eduGAIN identity federations, based on different authentication technologies. Joint Supercomputer Center of Russian Academy of Sciences (JSCC RAS) is participant of the above projects. JSCC RAS works on creation of identity federation in the Russian research and education community and supports coop-eration with European projects. JSCC supports Eduroam in Russia that is open for participation of any research and education institution.

3. Users group access rights in eduroam – federated user identity management system For research and educational networks [№4 за 2012 год]
Authors: Овсянников А.П., Овсянникова Т.В., Овчаренко С.А.
Visitors: 7757
The paper describes a federated identity management infrastructure based on eduroam. This technology enables secure authentication using single netid for network and resources access in eduroam federation. Major protocols and technologies for transparent user authentication are covered. A way of authorization, based on membership in institutional groups and individual user membership is proposed. For user authentication a service provider sends an authentication request contained the encrypted user name and password to user's institute RADIUS server (identity provider). Identity provider is determined by the domain user name/ The authentication request is passed through th eduroam hierarchy of proxy RADIUS servers. If the service provider provides special access for a certain group of users, it also sends a request to group identity RADIUS-server. A request passes through a hierarchy of group RADIUS servers for group membership checking. Eduroam federation and group RADIUS servers hierarchies are based on the domain name system. The implementation of these mechanisms requires a slight modification of service provider RADIUS server for group support and do not require changes of the identity provider and eduroam federations RADIUS servers. Group support is fully compatible with the existing eduroam infrastucture, the both types of RADIUS servers with and without group support can operate simultaneously.

4. An efficient application mapping algorithm for multiprocessor systems [№4 за 2012 год]
Authors: Киселёв Е.А., Аладышев О.С.
Visitors: 6738
This article describes a new application mapping approach for multiprocessor systems based on simulated annealing algorithm. The authors propose a model of multiprocessor system, which takes into account the heterogeneity of computing and communication resources, as well as a model of a parallel program based on the identification of typical communication operations between theprogram threads.The authors propose a parallel implementation of the algorithm simulation annealing to improve the quality of application mapping on the resources of multiprocessor computer system.The authors investigated the effect of competition in the network at the application work time.

5. Tuning execution of parallel programs [№4 за 2012 год]
Author: P.N. Telegin
Visitors: 6654
The article describes issues related to development of optimization methods for tuning programs for distributed memory parallel computers. Optimization tuning of parallel programs for given hardware is discussed. Tuning means choice of parameters of the parallel program for specific hardware. Parameters are the scheme of parallel execution and distribution of work among processors or cores. We consider execution of program with pre-made parallel decomposition and determined pseudolinear sections, including simple operation, branching structure, loops and unstructured regions with a single input and output. Program decomposition and its execution in flow, dynamic and static schemes is described. Three ways to estimate execution time of program fragments are investigated: time prediction, profiling, user assertion. Effect of magnification in predicting execution time of program fragments is described. The ratio of communication time operations time in processors is large, thus a careful program analysis to make a decision on parallel execution is required. Scheduling parallel loops is described. Formulas for efficiency estimation of the loops executions for different schemes and different transferred data sets are presented. The described method performance estimations was implemented in the automatic parallelization tool Ratio. Parallel program utilizing dynamic scheme was built using by Ratio parallelization tool and executed on MVS-100K supercomputer. The comparison of predicted and actual acceleration for application program is given.

6. Scientific Heritage of Russia Digital Library: Construction and sources aggregation philosophy [№4 за 2012 год]
Authors: Калёнов Н.Е., Савин Г.И., Serebryakov V.A., A.N. Sotnikov
Visitors: 13866
A construction philosophy for digital information resources by digital libraries tools is considered. Russian Scientific Heritage Digital Library Project is under consideration. Place and role of the Russian Scientific Heritage Digital Library among similar local and international projects where defined. Basic architecture statements and data aggregation tools for digital library construction were presented. Sources, types and structure of the digital objects were discussed. Data integration and federation concept as a main stone for the common education and scientific space is presented. Collaboration across national boundaries and information sources defined as imperative.

7. System dynamics in monitoring for mass vaccination of population [№4 за 2012 год]
Author: S.Yu. Lapshina
Visitors: 7844
The simulation modeling prototype of aggregate model with which help probably to organize monitoring of actions for vaccination of the population of quarantine zones of region in distribution and attenuation of mass infectious diseases is presented. The model is based on ideas and methods System Dynamics, and also on possibilities of widely known personified software package of simulation modeling IThink. At modeling accessible materials on flash and distribution in Moscow flu epidemics A/H1N1 in 2009–2010 were used. At the initial stage of prototyping following aggregate entities (resources) of model are distinguished: unvaccinated, vaccinated, successfully vaccinated, infected, sick, recover from, have acquired immunity, deaths. The current state of resources in the dynamics of the model describes the state of the situation as a whole. Resources (as primary basic concepts) are concentrated in interacting conceptual blocks vaccination and epidemic outbreaks, infection and disease progression. Behavior of the model considered in the context of plausible assumptions drawn from domestic and foreign sources. The current prototype has provided a series of simulation experiments to verify the model, confirming the integrity and objectivity of the results.

8. Social network of RAS – unified information environment for scientists [№4 за 2012 год]
Author: R.S. Suleymanov
Visitors: 8739
This article is about the modern tools of Internet communication, particularly about social networks. The main directions are: analysis of existing social networks, the story of their origin, their internal structure and operating principles. Described the main tools and techniques of prototyping and design of high loaded information systems. Also considered is the basic structure of RAS institutes and analyzed the problems of rapid exchange of information between researchers of RAS. Author writes of the design and development of a new social network for professional purposes. It is social network of the Russian Academy of Sciences for the science employee. Outlines the key objectives of the project and perspectives of the project implementation. The rationale for the choice of technologies used, examples of interfaces, collaboration tools and a review of the primary functional formed social network.

9. Branches optimization in binary translator for «Elbrus» architecture [№4 за 2012 год]
Author: A.A. Rybakov
Visitors: 4987
MCST company develops «Elbrus» architecture microprocessors. For «Elbrus» architecture binary translation system LIntel is developed. Lintel allows execute Intel x86 applications on «Elbrus» microprocessors. Multilevel optimizing binary translator is important component of Lintel. Branch preparing instructions in «Elbrus» architecture allow parallelize program execution and process branches instructions immediately, without loss of processor time. Applying branches optimization in binary translator allows transfer branch preparing instructions between linear sections of program. This optimization leads to system performance improvement.

10. Parallel software package for optimal development Of dynamic flow network [№4 за 2012 год]
Authors: Григоренко Н.Л., Жарков А.В., Пивоварчук Д.Г., Popova N.N.
Visitors: 6692
In this paper the model of the dynamic transport network is suggested. The dynamic transport network is defined as the delivery network which connects some amount of supplies, distributers and consumers. Parameters of this network (demand and consumption) stochastically change depending on the time. In this paper the problem of optimal development of such network is researched. Optimal development is the development, the object of which is the maximum satisfaction of consumers’ demand over the model time. The parallel algorithm of optimal development network problem solving is suggested. The necessity of creation such an algorithm is conditioned by large amount of computational operations, which are needed for problem solving of real transport network graphs. The suggested algorithm is taken as a basis of the software package, which allows solving the given problem at supercomputers. The ADOT language, which is the extension of DOT language, is described. This language allows representing the input data in user-friendly form. The approbation of this software package is executed at Lomonosov supercomputer. The graphics of algorithm time solving, speedup and efficiency are represented. The scheme of optimal task scheduling development is described in this paper.

| 1 | 2 | 3 | 4 | 5 | 6 | Next →