Research Events

Workshops

SDMAS’08 is an international workshop held within the 2008 International Conference on Parallel and Distributed Processing Techniques and Applications.

Second International Workshop on High Performance I/O Systems and Data Intensive Computing held within IEEE Cluster 2008.

Seminars

 

Title Speaker Date
Scalable Data Analysis on Parallel and Cloud Computing Systems Prof. Domenico Talia, Full Professor of Computer Engineering at University of Calabria 18, 19 y 20 de febrero de 2015
Abstract
 

Título: Scalable Data Analysis on Parallel and Cloud Computing Systems

Summary:
The analysis of large and distributed data repositories requires the combined use of smart data analysis techniques and scalable architectures to find and extract useful information from them. Parallel systems, Grids and Cloud computing platforms offer an effective support for addressing both the computational and data storage needs of Big Data mining and parallel analytics applications. In fact, complex data mining tasks involve data- and compute-intensive algorithms that require large storage facilities together with high performance processors to get results in suitable times. In this seminar we introduce the most relevant topics and the main research issues in high performance data mining including parallel data mining strategies, knowledge Grids and Cloud data mining. We also present some data mining frameworks designed for developing distributed data analytics applications as workflows of services on Grids and Clouds. In these environment data sets, analysis tools, data mining algorithms and knowledge models are implemented as single services that are combined through a visual programming interface in distributed workflows. Some applications will be also discussed.

Short Bio:
Domenico Talia is a full professor of computer engineering at the University of Calabria in Italy. He is a partner of two startups, Exeura and DtoK Lab. His research interests include parallel and distributed data mining, cloud computing, Grid services, knowledge discovery, mobile computing, green computing systems, peer-to-peer systems, and parallel computing. Talia published ten books and more than 300 papers in archival journals such as CACM, Computer, IEEE TKDE, IEEE TSE, IEEE TSMC-B, IEEE Micro, ACM Computing Surveys, FGCS, Parallel Computing, IEEE Internet Computing and international conference proceedings. He is a member of the editorial boards of IEEE Transactions on Computers, IEEE Transactions on Cloud Computing, the Future Generation Computer Systems journal, the International Journal on Web and Grid Services, the Scalable Computing: Practice and Experience journal, MultiAgent and Grid Systems: An International Journal, International Journal of Web and Grid Services, and the Web Intelligence and Agent Systems International journal. Talia served as a chair, organizer, or program committee member of several international conferences and gave many invited talks and seminars in conferences and schools. He is a member of the ACM and the IEEE Computer Society.

Links:

 

 

 

Title Speaker Date
Cloud Computing: Oportunidades de Innovación y Gestión Estratégica Profesor Ignacio Martín Llorente (DACYA, Facultad de Informática, Universidad Complutense de Madrid) 10 y 11 de Marzo
Abstract
 

Subject/T�tulo: Cloud Computing: Oportunidades de Innovaci�n y Gesti�n Estrat�gica
Professor/Profesor: Ignacio Mart�n Llorente
Dates/Fechas: 10 y 11 de marzo de 2014
Place/Lugar: Universidad Carlos III de Madrid Campus de Legan�s. Aula 2.3.A03

Abstract/Resumen:

Cloud Computing es un nuevo modelo de provisi�n de recursos como servicio, bajo demanda, y de forma el�stica y din�mica a trav�s de Internet. El objetivo de la conferencia es describir las posibilidades del Cloud Computing y el impacto potencial de su adopci�n como fuente de innovaci�n y ventaja competitiva. Este innovador modelo puede tener adem�s un impacto significativo en la estructura de costes de la empresa en funci�n de la variabilidad de la demanda de recursos. En el seminario se presentan las diferentes perspectivas y los beneficios del Cloud Computing, as� como los tipos de servicio Cloud y modelos de despliegue mostrando ejemplos de servicios reales. Adem�s se describen las consideraciones m�s importantes a tener en cuenta cuando se eval�a la adopci�n del modelo, incluyendo los principales obst�culos de adopci�n y lo factores a considerar en la negociaci�n con el proveedor. Por �ltimo, se discute la relevancia estrat�gica de la adopci�n del modelo Cloud Computing.

Contenidos:

  • Modelos de Provisi�n
  • Consideraciones sobre Adopci�n
  • El Factor Econ�mico
  • Gesti�n Estrat�gica
  • Modelos de Despliegue
  • El Futuro del Cloud

 

 

 

Title Speaker Date
Big data y Computación Cloud Profesora Maria de los Santos Perez (Universidad Politécnica de Madrid) 5/2/2013 y 6/2/2013
Abstract
 

Subject/T�tulo: Big data y Computaci�n Cloud
Professor/Profesor: Maria de los Santos Perez
Dates/Fechas: 5 y 6 de febrero de 2013
Place/Lugar: Universidad Carlos III de Madrid Campus de Legan�s. Aula 2.1.C01

Abstract/Resumen:

Actualmente se ha producido una gran explosi�n de datos digitales, debido principalmente a dos factores.

  • Por un lado, el inter�s de las empresas en tratar datos de diversa naturaleza, destacando de manera notoria datos de personas f�sicas o datos recogidos a partir de la observaci�n y el uso de sensores en diferentes entornos; y por otro lado, el relevante papel del ciudadano en la publicaci�n de contenidos, a trav�s de diferentes canales, como pueden ser dispositivos m�viles o redes sociales. A esta gran colecci�n de datos se le ha dado a conocer como Big data. El almacenamiento, tratamiento, an�lisis y visualizaci�n de Big Data supone un enorme reto que est� siendo abordado desde diferentes perspectivas.
  • Por otro lado, en la �ltima d�cada, el mundo de la computaci�n ha sufrido grandes cambios tecnol�gicos al aparecer un conjunto de iniciativas de diversas empresas fuertes del sector, como Amazon, Google, Salesforce o Microsoft, que ofrecen infraestructuras como servicio (IaaS, Intrastructure as a Service), plataformas como servicio (PaaS, Platform as a Service) y software como servicio (SaaS, Software as a Service), conceptos a los que junto con otras caracter�sticas se
    les ha dado el nombre gen�rico de computaci�n cloud (en nube) o cloud computing. Este nuevo paradigma implica un cambio conceptual en el uso y la gesti�n de los recursos inform�ticos, en el que dichos recursos constituyen el servicio b�sico, de manera an�loga al consumo por parte de los usuarios de servicios b�sicos como pueden ser el acceso a la red el�ctrica o a la red telef�nica.

    Este seminario describir� los retos de Big Data y las soluciones propuestas dentro de la computaci�n cloud. 

 

 

Title Speaker Date
Cloud Computing challenges: from a developer to a provider point of view Dra. Dana Petcu (Universidad de Vest) 29/03/2013 y 30/03/2013
Abstract
 

Developing an application that consumes resources from Cloud providers
is a challenge as a variety of tools and API are currently available,
most of them leading to a vendor lock-in of the developed application.
To avoid this problem several considerable efforts are devoted by researchers and companies.
Focus will be put on current solutions to overcome this status.
The presentation will include:

  1. generalities about Cloud computing (see for example [1])
  2. a training in mOSAIC technologies (see for example [2,3])
  3. overview of on-going activities on Cloud Computing in European Commission projects
    (see for example [4,5]), with accent on model-driven engineering & Clouds (MODAClouds [6])
  4. hints about the technologies at the provider side: scheduling and auto-scaling (AMICAS [7])

Useful links:

  1. Lecture about Cloud computing (http://web.info.uvt.ro/~petcu/distrib/TDS12.pdf)
  2. Portability tutorial (http://%20http//www.iaria.org/conferences2012/filesCLOUDCOMPUTING12/portability_tutorial.pdf)
  3. mOSAIC developers (http://developers.mosaic-cloud.eu/)
  4. Cloud computing in ERA (http://www.c-s-p.org/Flyers/978-1-4438-3507-7-sample.pdf)
  5. Advances in Clouds (http://cordis.europa.eu/fp7/ict/ssai/docs/future-cc-2may-finalreport-experts.pdf)
  6. MODAClouds (http://www.modaclouds.eu/)
  7. AMICAS (http://amicas.hpc.uvt.ro/)

 

 

 

Title Speaker Date
High Performance Multimedia Computing Raffaele Montella, Universidad Parthenope – Napoli, Italia 22, 23 y 25 de mayo de 2012
Abstract
 

This course is concerned with the high performance computing tools for
multimedia applications historically ranging from the cluster
ray�‐trace rendering of early 80�s to the today strict requirements
for near real time applications walking throw the amazing techniques
of elastic transcoding needed by the upcoming mobile devices provided
by tridimensional cameras.

The high performance computing usually
referred as HPC is provided by a floating formal definition that could
sound as �the most up�-to-date technology achieving the best computing
results�. Thanks to this philosophically-�powered definition we could
state the Babbage machine, if ever build, was the Victorian equivalent
of today BlueGene. The technology related to the HPC field changed
many times in the last few years: symmetric multi processors,
distributed memory computing clusters, clusters of many-�‐core
provided nodes, general purpose GPUs, hieratical parallelism, grid and
cloud computing. Stepping over the hype of each technology today is
more difficult explain what is not HPC thank what it is.

In the field
of multimedia application the HPC technologies play the same role of
special effects in movies: as better as not visible. Todays HPC
technologies permits the near real time responsiveness needed in the
business of multimedia applications. The rise of the pervasive and
ubiquitous access to multimedia contents and the need to contribute to
them increased the need of distributed high performance multimedia
computing.

 

 

Title Speaker Date
Harnessing Massively Parallel Processors Matei Ripeanu, Universidad British Columbia, Canada 29 de mayo, 31 de mayo y 1 de junio
Abstract
 

Virtually all semiconductor market domains, including PCs, game
consoles, mobile handsets, servers, and supercomputers, are converging
to concurrent platforms. There are two important reasons for this
trend. First, these concurrent processors can potentially offer more
effective use of chip space and power than traditional monolithic
microprocessors for many demanding applications. Second, an increasing
number of applications that traditionally used Application Specific
Integrated Circuits (ASICs) can be implemented using concurrent
processors to improve functionality and reduce engineering cost. The
real challenge is to develop applications software that effectively
uses these concurrent processors to achieve efficiency and performance
goals.

The aim of this course is to provide students with knowledge and
hands-on experience in developing applications massively parallel
processors as well, an understanding of current research topics in
this area, and an opportunity to write and critically review research
papers. Many commercial offerings from NVIDIA, AMD, and Intel already
offer concurrency at massive scale. Effectively programming these
processors requires in-depth knowledge about parallel programming
principles, as well as the parallelism models, communication models,
and resource limitations of these processors.

The course will cover fundamental issues related to architecting and
building applications that harness massively parallel processors, with
a focus on graphical processing units (GPUs): programming models for
massive parallelism, application optimization techniques, compile and
runtime support, as well as case studies for specific
application-domains.

 

 

Title Speaker Date
Distributed Systems, Grids and Clouds Prof. Helen D. Karatza Lunes 27 a Miercoles 29 de Febrero
Abstract
 

Subject/T�tulo: Distributed Systems, Grids and Clouds
Professor/Profesor: Helen D. Karatza
Dates/Fechas: 27, 28 y 29 de febrero de 2012
Place/Lugar: Universidad Carlos III de Madrid Campus de Legan�s. Aula 12C16

Abstract/Resumen:

Distributed systems are essential to meet the needs of demanding applications because they have increased computing power, are cost-effective and are easily scalable. They consist of loosely interconnected processors, where jobs to be processed are in some way apportioned among the processors, and various techniques are employed to coordinate processing.
Computational and data grids and clouds are large-scale distributed systems used for serving demanding jobs. Their performance became more important due to the tremendous increase of users and applications.
A grid is a decentralized heterogeneous system with its resources geographically distributed at different sites. The main idea of grid computing is to use a large number of distributed resources in order to solve complex and computationally demanding problems that practically could not be solved on a single resource.
Cloud computing evolves from grid computing. Clouds provide elasticity in resource usage as well as a pay as-you-go opportunity.
Because of the nature of these systems, there are important issues that must be addressed, such as: efficient scheduling, resource management, energy efficiency, reliability, security and trust, cost, availability, quality.
Resource allocation and scheduling is a difficult task in grids and clouds where there are many alternative heterogeneous computers. The performance evaluation of these systems is often possible only by simulation rather than by analytical techniques, due to the complexity of the systems. Simulation can provide important insights into the efficiency and tradeoffs of scheduling in large-scale complex distributed systems, such as grids and clouds.

Topics Covered:
* Distributed systems algorithms (distributed consensus, leader election)
* Real time distributed systems
* Resource management (discovery, allocation)
* Scheduling in grids and clouds
* Energy efficiency
* Performance � Modelling and Simulation

Biography/Sobre la ponente:

Helen Karatza is a Professor in the Department of Informatics at the Aristotle University of Thessaloniki, Greece. She received the B.S. degree and Ph.D. from the Aristotle University.
Dr. Karatza�s research interests include Computer Systems Modeling and Simulation, Performance Evaluation of Parallel and Distributed Systems, Grid and Cloud Computing, Resource Allocation and Scheduling and Real-time Distributed Systems.
Dr. Karatza has authored or co-authored over 165 technical papers and book chapters. She is Editor-in-Chief of the Elsevier Journal Simulation Modeling Practice and Theory

Materiales:

karatza

 

 

 

Title Speaker Date
Data Management in Wireless Sensor Networks Prof. Stefano Chessa (University of Pisa, Italy) 17 de Abril
Abstract
 

Wireless sensor networks (WSN) are a recent technology aimed at the automation of environmental monitoring by means of highly distributed, low-cost and wireless sensors. Some WSN paradigms provide a unified organization of WSN by combining database technologies with WSNs. In these paradigms the database to be queried is the physical environment where the WSN is deployed. These approaches abstract specific sensor features and offer to the applications an SQL-like query language through which it can program the network and access the sensed data. More recently the same approach has been extended to the tracking of composite, moving events.

Under this trend of research we identify data collection, intra-sensor and inter-sensor communication as recurring activities in WSN applications and propose a layered approach in which the lower layer models the communications and the sampling functions of the sensors in terms of streams, and the upper layer performs query processing on streams. The MaD-WiSe system is a wireless sensor network database based on this approach and designed to perform in-network distributed query processing. The talk describes the MaD-WiSe system, and discusses its extension for the tracking of moving events.

 

11:00-13:00h

 

 

Title Speaker Date
Trends and future of C++: Evolving a systems language for performance Prof. Bjarne Stroustrup (Texas A&M University) and
Michael Wong (IBM)
18th March 2011 + 28th March 2011
Abstract
 

T�tulo: Trends and future of C++: Evolving a systems language for performance
Profesor: Prof. Bjarne Stroustrup (Texas A&M University) and Michael Wong (IBM)
When: 18th March 2011 + 28th March 2011
Where: Universidad Carlos III de Madrid Campus de Legan�s. Sal�n de actos

Abstract:
Over the last decades, C++ has become one of the most widely used languages supporting object-oriented programming by making abstraction techniques affordable and manageable for mainstream projects. Using C++ as his tool, Prof. Bjarne Stroustrup pioneered the use of object-oriented and generic programming techniques in application areas where efficiency is a premium; examples include general systems programming, switching, simulation, graphics, user-interfaces, embedded systems, and scientific computation. The influence of C++ and the ideas it popularized are clearly visible far beyond the C++ community. Languages including C, C#, Java, and Fortran99 provide features pioneered for mainstream use by C++.

C++ is a general purpose programming language with a bias towards systems programming which supports multiple styles of programming. Since 1998 C++ is an ISO standard. In these days a new version of the language is about to become a new ISO standard. This seminar will start by introducing the new features of the language and showing the challenges of evolving an existing programming language with a huge code base. Additionally, the seminar will introduce the new concurrency model for C++ and its impact both on the language and the standard library. Finally, a set of tools for parallel software development (Intel Parallel Building Blocks) will be presented.

Links: http://www.arcos.inf.uc3m.es/~jdaniel/sem-cpp-11/

 

 

 

Title Speaker Date
Sensores inalámbricos Prof. Dr. Pedro José Marrón 8, 9 y 10 de febrero
Abstract
 

T�tulo: Wireless sensor networks
Profesor: Prof. Dr. Pedro Jos� Marr�n
When: 8-10 Feb.
Where: Universidad Carlos III de Madrid Campus de Legan�s. 21C10

Abstract:

Links:

 

 

 

Title Speaker Date
High-Performance Computing and Image Processing Hamid R. Arabnia 2-3, March
Abstract
 

T�tulo: High-Performance Computing and Image Processing
Profesor: Hamid R. Arabnia
When: 2-3 March.
Where: Universidad Carlos III de Madrid Campus de Legan�s.

Abstract:

The talk would include:

* an introduction to image processing
* an introduction to high-performance computing
* the integration of the above two for form applications
* (example aplications would be in medical imaging, security, face recognition, …

 

 

 

Title Speaker Date
Distributed Web-based architectures for content and service delivery Michele Colajanni, University of Modena and Reggio Emilia 6, 7 y 9 de febrero de 2007.
Abstract
 

Título: Distributed Web-based architectures for content and service delivery
Profesor: Michele Colajanni, University of Modena and Reggio Emilia
Fechas: 6, 7 y 9 de febrero de 2007.
Lugar: Universidad Carlos III de Madrid Campus de Legan�s. Aula 71J06

Resumen:

The successful of the Web in the last decade has significantly
modified the type of resources that are distributed through the Internet.
The initial static contents are now enriched by dynamically generated
resources, with an increasing amount of multimedia and personalized
services. This evolution is shifting the research focus from Web content
delivery to service delivery that introduces new scalability and
performance
problems to the hardware/software infrastructures that have to generate
and
distribute content. Indeed, the heterogeneity and richness of present and
future services increase the necessity of high performance architectures
that exploit caching and replication through multiple servers that are
distributed over a local and/or a geographical area. In these lessons, we
investigate the main architectures for efficient generation, adaptation
and
delivery of contents and services through the Internet with a major focus
on
performance. We will illustrate the advantages and limits of current
solutions, and possible directions for future research.

Sobre el ponente: Prof. Colajanni is Full Professor at Univeristy of Modenna
and Reggio Emilia. Previously he has performed research activities at
University of Rome and IBM T.J. Watson Research Center. He has published
about 120 publications in referred journals, book chapters and conference
proceedings (A list of the most recent papers can be found at:
http://weblab.ing.unimo.it/people/colajanni/publications.shtml). He is de
director of the Web Engineering and Performance Laboratory (WEBlab,
http://weblab.ing.unimo.it).

Material del seminario:

Part 1
Part 2
Part 3
Assignments

 

 

 

Title Speaker Date
Filecules and Small Worlds in Scientific Communities: Characteristics and Signif Adriana Iamnitchi, University of South Florida July 2007
Abstract
 

T�tulo:Filecules and Small Worlds in Scientific Communities: Characteristics and Signif
Profesor:Adriana Iamnitchi
Origen:University of South Florida
Fecha impartici�n: Julio de 2007.
Duraci�n: 10 horas.

Abstract:

Most of today’s science depends on the processing of massive amounts of data in multi-institutional and even international collaborations. Grid computing focuses on enabling resource sharing for wide-area collaborations and has currently reached the stage where deployments are mature and many collaborations run in production mode. As with any growing technology, Grid usage characteristics (that inherently affect performance) could not have been predicted before or during design and implementation. This lack of evidence in usage characteristics has three significant outcomes: (1) resource management solutions are evaluated on irrelevant traces; (2) quantitative comparison of alternative solutions to the same problem becomes impossible due to different experimental assumptions and independently generated workloads; and (3) solutions are designed in isolation, to fit the particular and possibly transitory needs of specific groups.

These concerns led us to analyze more than two years of workloads from a high-energy physics collaboration. In addition to contradicting previously accepted models, we discovered two novel data-usage patterns. First, a data-centric analysis reveals the existence of filecules, groups of files that are always processed together. Second, a user-centric analysis discovers small-world properties in data sharing that show emergent, interest-based grouping of users. We show that exploiting these patterns for designing resource management solutions leads to better scalability, lower costs, and increased adaptability to changing environments. In addition, these traces helped us evaluate the viability of offloading the storage needs of the high-energy physics collaboration to the Amazon Simple Storage Service (S3). We identify ways to exploit application-specific characteristics for reducing S3 storage costs and discuss the support that a storage utility service such as S3 needs to provide in order to support scientific data-sharing collaborations.

 

 

 

Title Speaker Date
Environmental Modeling using Grid Computing Raffaele Montella, University of Naples “Parthenope” June 2007
Abstract
 

T�tulo:Environmental Modeling using Grid Computing
Ponente:Raffaele Montella
Origen:University of Naples �Parthenope�
Fecha impartici�n: Junio de 2007.
Duraci�n: 10 horas.

Abstract:

The environmental modeling is one of the most attractive fields in the grid computing technologies based application development because it involves some specific needs suited to be solved by high performance computing, distributed computing, workflow management, service oriented architectures, network of shared instruments handling and so on.

The Department of Applied Sciences of the University of Naples �Parthenope� joined the historical knowledge in environmental modeling and applied mathematics with the strongly computer science oriented grid computing components and applications development field. In this contest a grid architecture was designed and developed spreading over it open source environmental numerical models building a virtual laboratory for meteo-marine forecasts, now casts and scenario analysis.

The developed application runs dynamically thanks to a from scratch developed job flow scheduler (JFSS, Job Flow Scheduler Service), resource broker (RBS: Resource Broker Service), five dimensional data distribution service (GDDS: Grads Data Distribution Service) and an instrument virtualization service (IS, Instrument Service) using Globus Toolkit version 4 web services. The actual grid application integrates the WRF (Weather Research and Forecast) forcing in one way mode both the Princeton Ocean Model (POM) and the WaveWatch III ocean wave model and the Weather Research and Forecast (WRF) model forcing the Comprehensive Air-Quality Model with Extensions (CAMx). The aim of this seminar suite is to provide a theoretical and practical skills on grid enabling environmental models using both the approaches based on remote execution via specific protocols as the Globus GT2 does and a service oriented architecture where different models are coupled thanks the GT4 provided web services.

 

 

 

Title Speaker Date
Ubiquitous and pervasive Web: services and architectures Riccardo Lancellotti University of Modena e Reggio Emilia April 2007
Abstract
 

Título:Ubiquitous and pervasive Web: services and architectures

Riccardo Lancellotti

University of Modena e Reggio Emilia

Fecha impartici�n: Abril de 2007.

Duraci�n: 10 horas.

Abstract:

The evolution of network-based services will be characterized by mobile users accessing highly complex, multimedia-oriented and personalized services. Access to content, information and services will be tailored to the client device, the user preferences and its context. The so-called social-networking services represent an example of the next generation of network-based services.

The course will provide a description of possible future scenarios, describing the currently available services and outlining their possible evolution. Furthermore, the course will discuss how this evolution is likely to impact on the server infrastructure. To this aim we will present several systems and architectures proposed by industry and in scientific literature, describing their characteristics and heir performance. We will discuss how these proposal can address the need of next generation network-based services.

 

 

 

Title Speaker Date
TÉCNICAS DE ALTO RENDIMIENTO EN EL DISEÑO DE PROCESADORES Prof. Francisco Tirado, Universidad Complutense de Madrid, España. 27 al 29 de marzo 2006
Abstract
 

Título:T�CNICAS DE ALTO RENDIMIENTO EN EL DISE�O DE PROCESADORES
Ponente: Profesor Francisco Tirado
Universidad Complutense de Madrid, Espa�a.
Fecha impartici�n: 27 al 29 de marzo 2006.
Horario: de 16 a 20 horas. Duraci�n: 10 horas.
Sala: Adoraci�n de Miguel. 12 C 16
Escuela Polit�cnica Superior. Universidad Carlos III. Legan�s, Madrid

Material del seminario:

carlosiii-1.zip

MATERIAL COMPLETO

Abstract:

Objetivos:

Revisar las t�cnicas actuales de dise�o de procesadores, dando una visi�n integrada de las interdependencias entre la evoluci�n de la tecnolog�a y la arquitectura de estos procesadores integrados
Conocimientos y destrezas que se requieren: Arquitectura de Computadores: nivel avanzado;

Contenidos:

1. Introducci�n: Tendencias tecnol�gicas, costo/ rendimiento . ProCesadores superescalares . Tratamiento de dependencias de control . Acceso a memoria, limites del modelo . Predicci�n de datos Arquitecturas VLIW Compilaci�n y Optimizaci�n Dinamica
2. Tratamiento de dependencias de control I
2.1 Introducci�n
2.2 Tratamiento de saltos
2.3 T�cnicas de predicci�n de saltos
3. Tratamiento de dependencias de control II
3.1. Predicci�n multiple
3.2. M�s AB en la etapa fetch: Branch Address Cache, Collapsing Buffer, Trace Cache, Multiple Block Ahead Branch Predictor, Path Prediction
4 Acceso a memoria, limites del modelo
4.1 Acceso a Memoria: Prebusqueda SW, Prebusqueda HW, Caches sin bloqueo, Especulaci�n de Load
4.2. La siguiente generaci�n: Procesadores multithreading, Procesadores de traza, Procesadores superespeculativos, Procesadores multiescalares, Multiprocesadores en un chip
5. Manejo del flujo de datos
5.1. Localidad de datos
5.2. T�cnicas no especulativas
5.3. T�cnicas especulativas
6. Optimizaci�n y traducci�n binaria din�micas

Bibliograf�a:

Superscalar Microprocessor Design, M. Johnson, Prentice Hall, 1991
Advanced Computer Architectures, D. Sima, T. Fountain, P. Kacsuk, Addison- Wesley, 1997
The Anatomy of a High- Performance Microprocessor A Systems Perspective, B. Shriver, B. Smith, IEEE Computer Society, 1998
Modern Processor Design; J.P. Shen, M. H. Lipasti, ; McGraw Hill, 2005;

 

 

 

Title Speaker Date
Ingeniería de Requisistos: conceptos, procesos y estado de la investigación Pere Botella Catedrático de Lenguajes y Sistemas Informáticos Universidad Poli 1, 2 y 3 de febrero de 2005
Abstract
 

T�tulo:INGENIER�A DE REQUISITOS: CONCEPTOS, PROCESOS Y ESTADO DE LA INVESTIGACI�N

Ponente:Prof. Pere Botella
Catedr�tico de Lenguajes y Sistemas Inform�ticos
Universidad Polit�cnica de Catalu�a
Fecha impartici�n: 1, 2 y 3 de febrero de 2005.
Horario: de 16 a 20 horas. Duraci�n: 10 horas.
Sala: 23 A 02. Edificio Sabatini.

Escuela Polit�cnica Superior. Universidad Carlos III. Legan�s, Madrid

 

Material del seminario:

ingreqciii.zip

Ejemplo pr�ctico

Abstract:

El seminario tiene dos partes diferenciadas: la primera (6 primeros temas) repasan los diferentes aspectos y actividades de la Ingenier�a de Requisitos, y la segunda (3 �ltimos temas) se centra en los temas de investigaci�n actuales, los grupos que la llevan a cabo y los trabajos del grupo al que pertenece el profesor. La primera parte incluye peque�os ejercicios pr�cticos.

 

PROGRAMA:

  • Definiciones y �mbito
  • Los procesos: conceptos b�sicos
  • El proceso de la Ingenier�a de Requisitos: modelos de proceso
  • Las actividades: descripci�n de las diferentes tareas
  • Propiedades y validaci�n
  • Gesti�n de Requisitos
  • La investigaci�n: visi�n general
  • Grupos de investigaci�n relevantes y sus l�neas
  • La investigaci�n en el grupo GESSI

BREVE CV DEL PONENTE:

El Dr. Pere Botella es Catedr�tico en la UPC (Universitat Polit�cnica de Catalunya). Ha trabajado activamente en el campo de la ingenier�a del software desde hace m�s de 25 a�os. Ha sido Decano de la Facultad de Inform�tica (1992-1998) y Vicerrector de la UPC (1982-1986 y 1998-2002). Desde 1989 a la actualidad, director del Master en Ingenier�a del Software de la UPC. Autor de mas de 40 publicaciones. Miembro del comit� de programa de diversas conferencias internacionales, incluyendo ESEC, ICSE, RE, ICCBSS, SEKE y otras, habiendo sido �executive chair� y co-editor de ESEC�95. Miembro de los comit�s permanentes de ESEC y de JISBD. Ha sido coordinador para Espa�a de la red RENOIR (European Network of Excellence in Requirements Engineering).

 

 

 

Title Speaker Date
Tecnologías de Máquinas Virtuales Juan Manuel Cueva. 17 al 19 de enero de 2005.
Abstract
 

Título:Tecnolog�as de M�quinas Virtuales

Profesor: Juan Manuel Cueva

Fechas: 17 al 19 de enero de 2005

Lugar: Universidad de Oviedo

Material del seminario:

material.zip

 

 

 

Title Speaker Date
Architecture/Software Co-Design Techniques For Software Security And Data Intens Alok Choudhary 9 y 10 de diciembre de 2004.
Abstract
 

T�tulo: ARCHITECTURE/SOFTWARE CO-DESIGN TECHNIQUES FOR SOFTWARE SECURITY AND DATA INTENSIVE COMPUTING
Ponente:Prof. Alok Choudhary
Department of Electrical Computing Engineering
Nothwestern University
Chicago, EE.UU.
Fecha impartici�n: 9 y 10 de diciembre de 2004.
Horario: de 15 a 20 horas.
Duraci�n: 10 horas.
Lugar: Escuela Polit�cnica Superior, Universidad Carlos III. Legan�s, Madrid
Edificio Sabatini. Sala: 21C01

Material del seminario:

alok-2004.zip

Abstract:

This seminar shows the new in trends in Architecture/Software Co-Design Techniques for Software Security and Protection and Data Intensive Computing and Ultra-Scalability for Parallel I/O. The objective of this presentation is to present various recent results and approaches to security and protection of systems. In particular, this presentation will focus on architectural advances as well as co-design and compilation approaches to enhance software protection and security. Moreover, it will present data intensive computing as required in scientific data analysis, management and discovery.

Topics:

1. Introduction

2. Security Life-Cycle

3. Threat Models

4. Architectural approaches (Intel LeGrande, ARM Trustzone etc.)

5. Trusted Computing Group’s specification

6. HW/SW co-design techniques and security performance tradeoffs.

7. Parallel File Systems: new approaches to scalability for ultrascale computing

8. Ultra-scalability using access pattern analysis and elimination of synchronization

9. Scalable runtime systems for parallel I/O

10. Scalable data mining techniques and recent results.

11. Conclusions.

Short biography of the author:

Prof. Alok Choudhary got his Ph.D. at the University of Illinois, Urbana-Champaign, in 989. He is currently a full Professor, ECE Department in the Northwestern University, Chicago, USA. He has got several awards, including the NSF Young Investigator Award, in 1993, the IBM Faculty Development Award, in 1994, the Intel Faculty Research Award, in 1993, and the IEEE Engineering Foundation Award, in 1990. His research interest includes compilers and Runtime Systems for High-Performance, embedded and adaptive computing systems and power-aware systems, High-Performance Databases, OLAP and datamining and Parallel and distributed Input-Output for Scientific and Information Processing Applications.

 

 

 

Title Speaker Date
Grid Computing for Data Intensive Applications Heinz Stockinger CERN, European Organization for Nuclear Research Ginebra, Sui 15 a 17 Marzo de 2004
Abstract
 

Título:Grid Computing for Data Intensive Applications

Dr. Heinz Stockinger

CERN, European Organization for Nuclear Research

Ginebra, Suiza

15 a 17 Marzo de 2004

Horario: de 16 a 20 horas.

Aula de audiovisuales. Biblioteca. Planta s�tano

Duraci�n: 10 horas.

Lugar: Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

Material del seminario:

gridseminar.zip

 

Breve biograf�a del autor:

Dr. Heinz Stockinger holds a Ph.D. in ‘Computer Science and Business Administration’ from the

University of Vienna, Austria. He has a Fellow position at CERN (European Organization for

Nuclear Research, Geneva, Switzerland) where he is working in the EU DataGrid (EDG) project. His

research interest is Grid Computing and in particular data management in Data Grids. In the EU

DataGrid project he is responsible for replication software. In addition, he is EDG’s Education &

Outreach Manager and thus teaching of Grid Computing.

In the last few years Grid computing has become very popular in several communities in research

as well is in industry and commerce. Whereas classical Computing Grids deal with CPU intensive

applications, Data Grids are suitable to deal with large amounts of data and thus can support data

intensive applications.

The seminar covers several aspects of Grid computing, provides a basic introduction to Data Grids

and state of the art in this research domain. Emphasis will be put on typical data management problems

of data intensive applications and how there are approached in currently existing Grid systems. As an

example Grid, there is a focus on the European DataGrid project with its current software systems.

Additional topics will cover specific middleware issues in job submission, data management, information

systems as well as on DataGrid’s security architecture. The seminar closes with a short outlook into

the future of Grid technologies.

Temario:

*Introduction to Grid Computing

*The European DataGrid Project and Other Grid Projects in Europe

*Grid Architecture and Services

*Data and Replica Management

*Selected Topics on Grid Services

Requirements:

Students attending the seminar should have a basic knowledge of:

-the Unix/Linux operation system

-distributed computing

 

 

 

Title Speaker Date
Scalable Distributed Data Structures : State-of-the Art Witold Litwin 25-27 de febrero de 2004
Abstract
 

Título:Scalable Distributed Data Structures: State-of-the Art

Prof. Dr. Witold Litwin

Witold Litwin

ACM Fellow

Professeur U. Paris 9 Dauphine

Directeur du CERIA

 

Fecha impartici�n: 25-27 de febrero de 2004.

Horario: de 16 a 20 horas.

Aula de audiovisuales. Biblioteca. Planta s�tano

Duraci�n: 10 horas.

Lugar: Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

 

Material del seminario:

litwin.zip

 

Breve biograf�a del coordinador:

Dr. Witold A. Litwin is Professor of Computer Science at the University Paris 9 (Dauphine) since 1990. He is also the Director of Centre d’Etudes et de Recherches en Informatique Appliqu�e (CERIA) at U. Paris 9. His research areas are the multidatabase systems and the data structures, the scalable distributed data structures especially. The multidatabase system design techniques and the linear hash data structures he has proposed in 80-ties are among most renowned contributions to these domains. They are routinely taught in the database and the data structures courses. They are also present in major database systems, as well as in other popular products, e.g., of Netscape and of Microsoft. Dr. Litwin has been invited lecturer and scientist at several universities. In the US, he taught database courses among others at UC Berkeley in 1992-94, at Santa Clara University in 1991 and Stanford University in 1990-91. His course on Multidatabase Systems was broadcast on Stanford Educational TV Network (SITN). He was visiting scientist at prominent US research centers, including IBM Almaden Research Center, in 1997 and 1998 and part time at Hewlett Packard Palo Alto Laboratories, between 1990 and 1994, as well as at U. of Maryland in 1989. Between 1980 and 1989 Dr. Litwin was Research Director at Institut National de Recherche en Informatique et Automatique (INRIA, France), and Expert for ISO Committee on Open Systems. Dr. Litwin wrote over hundred fifty research papers, edited or contributed to eleven books, and was Program Committee member of over fifty international database conferences. He is ACM-Fellow 2002, and member of IEEE.

Abstract:

SDDSs are a new class of data structures specifically for multicomputers. An SDDS file can span over the storage of many computers linked through a high-speed network. Such files typically reside for processing in the distributed RAM storage. They can attain sizes and access performance impossible for the traditional files. These capabilities of SDDSs open new perspectives for many applications : DBMSs, multimedia and video servers, real-time or high performance computing�

Temas:

The SD linear hash partitioning schemes

LH* The SD range partitioning schemes RP*

The SD high-availability schemes: LH*rs & others

Overview of other notable SDDS schemes

The algebraic signatures in SDDSs

The SDDS-2000 architecture and demo

The SD-SQL Server SDDS applications & concluding remarks

 

 

 

Title Speaker Date
Cómputo Paralelo: Aplicaciones, Arquitecturas y Programación Jorge Buenabad 25 al 27 de noviembre de 2003
Abstract
 

Título:C�mputo Paralelo: Aplicaciones, Arquitecturas y Programaci�n

Prof. Dr. Jorge Buenabad

Secci�n de Computaci�n

Departamento de Ingenier�a El�ctrica

Centro de Investigaci�n y de Estudios Avanzados del IPN

Fecha impartici�n: 25 al 27 de noviembre de 2003.

Horario: de 16 a 20 horas.

Duraci�n: 10 horas.

Lugar: Escuela Polit�cnica Superior. Aula 20D14.

Universidad Carlos III.

Legan�s, Madrid

Material del seminario:

buenabad.zip

 

Breve biograf�a del autor:

Ingeniero por la Universidad Aut�noma de M�xico DF. Doctor en Ciencias (Ingenier�a El�ctrica, 1998) Universidad de Bristol, Inglaterra.
Imparte docencia e investiga en temas relacionados con la Arquitectura e Ingenier�a de Computadores, Arquitecturas no Cl�sicas y Redes de Altas Prestaciones.

Sus temas de investigaci�n en la actualidad incluyen : Sistemas operativos, arquitectura de computadoras, lenguajes de programaci�n de sistemas en paralelos y distribuidos.

Abstract:

En este seminario se presentar� una vista general de las aplicaciones y de los mecanismos que hacen posible el c�mputo paralelo. Se comienza para ello con la justificaci�n del problema y de la necesidad de soluciones, para comentar a continuaci�n las soluciones tecnol�gicas existentes y los paradigmas de programaci�n paralela m�s adecuados para cada tecnolog�a y problema. Se presenta arquitecturas escalables de memoria compartida y las requerimientos del software del sistema para administrarlas eficazmente. Se presenta el hardware y software utilizado para desarrollar y correr aplicaciones paralelas usando mensajes. Se presenta el hardware y software utilizado para desarrollar aplicaciones de memoria compartida. Se presenta la motivaci�n para utilizar c�mulos de computadoras para aplicaciones paralelas y, en particular, para utilizarlas bajo el modelo de memoria compartida.

Temas:

1.Introducci�n

1.1.Aplicaciones que requieren paralelismo.

1.2.Computadoras SIMD y MIMD (Single Instruction Multiple Data y
Multiple Instruction Multiple Data).

1.3.Modelos de Programaci�n MIMD: paso de mensajes, memoria
compartida.

1.4.Dise�o de aplicaciones paralelas.

1.5.Herramientas y ambientes.

2. Sistemas de Paso de Mensajes

2.1.Multicomputadoras, c�mulos de computadoras.

2.2.Software del sistema.

2.3.Paso de mensajes en maquinas de memoria compartida.

2.4.Ventajas y desventajas.

3.Sistemas de Memoria Compartida

3.1.Motivacion: facilidad de programacion y portabilidad de las
aplicaciones.

3.2.Hardware: multiprocesadores

3.3.Consistencia de datos.

3.4.Software del sistema: lenguajes, librerias, sistema operativo.

3.5.Desventajas: no son escalables.

4.Multiprocesadores Escalables.

4.1.Maquinas NUMA (non-uniform memory access machines).

4.2.Maquinas CC-NUMA (cache coherente-NUMA)

4.3.Requerimientos del software del sistema para administrar
NUMAs y CC-NUMAs.

4.4.Arquitecturas de Difusion de Datos.

4.5.Requerimientos del software del sistema para administrar
ADDs.

5.Memoria Compartida en Cumulos de Computadoras.

5.1.Motivacion: bajo costo, prop�sito general, f�cil actualizaci�n
del hardware.

5.2.Herramientas de desarrollo.

5.3.Enfoques para soportar memoria compartida.

5.4.Ventajas y desventajas.

5.5.Tendencias actuales.

 

 

 

Title Speaker Date
Técnicas para la mejora de las prestaciones y servicios ofrecidos por la red Inf Prof. Dr. José Duato Prof. Dr. Antonio Robles 16 al 18 de diciembre de 2003
Abstract
 

T�tulo:T�cnicas para la mejora de las prestaciones y servicios ofrecidos por la red InfiniBand

Ponentes:Prof. Dr. Jos� Duato
Prof. Dr. Antonio Robles
Depto. de Inform�tica de Sistemas y Computadores (DISCA)
Escuela T�cnica Superior de Inform�tica Aplicada
Universidad Polit�cnica de Valencia

Fecha impartici�n:16 al 18 de diciembre de 2003.
Horario: de 16 a 20 horas.
Duraci�n: 10 horas.
Lugar: Aula 20D14, Escuela Polit�cnica Superior, Universidad Carlos III, Legan�s, Madrid.

Material del seminario:

slides-duato.zip

 

Breve biograf�a del coordinador:

Jose Duato received the MS and PhD degrees in electrical engineering from the Technical University of Valencia, Spain, in 1981 and 1985, respectively. In 1981 he joined the Department of Systems Engineering, Computers and Automation at the Technical University of Valencia. Currently, Dr. Duato is Professor in the Department of Computer Engineering (DISCA). He was also an adjunct professor in the Department of Computer and Information Science, The Ohio State University.

His current research interests include interconnection networks, multiprocessor architectures, networks of workstations, and switch fabrics for IP routers. Prof. Duato has published over 230 refereed papers. He proposed the first theory of deadlock-free adaptive routing for wormhole networks. This theory has been highly cited and used by other researchers. In particular, versions of this theory have been used in the design of the routing algorithms for the MIT Reliable Router, the Cray T3E supercomputer, the internal router of the Alpha 21364 microprocessor, and the BlueGene/L supercomputer. Also, Myricom Inc. has defined a special ITB packet type for their Myrinet network, thus providing support for the �In-Transit Buffer (ITB)� routing techniques developed by Prof. Duato.

Abstract:

Este seminario pretende mostrar una panor�mica de la red Infiniband, sus algoritmos de encaminamiento, calidad de servicio, etc.

Temas:

Introducci�n

InfiniBand: Conceptos b�sicos

Encaminamiento en InfiniBand

Viabilidad del uso de encaminamiento up/down en InfiniBand

T�cnicas mejoradas de encaminamiento determinista

T�cnicas de encaminamiento adaptativo en InfiniBand

Calidad de servicio en InfiniBand

Clasificaci�n del tr�fico y garant�as de calidad de servicio

C�lculo de las tablas de arbitraje

Reconfiguraci�n de la red

T�cnicas mejoradas de reconfiguraci�n para InfiniBand

 

 

 

Title Speaker Date
Parallel I/O: Lessons learnt in the last 25 years Toni Cortes. 5, 6 y 7 de marzo de 2003
Abstract
 

Título:Parallel I/O: Lessons learnt in the last 25 years

Toni Cortes

Universidad Polit�cnica de Catalu�a.

Departamento de Arquitectura de Computadores

Fecha impartici�n: 5, 6 y 7 de marzo de 2003.

Horario: de 16 a 20 horas. Duraci�n: 10 horas.

Lugar: Aula 20D10 Edificio Sabatini
 

Responsable: Jes�s Carretero (jcarrete@arcos.inf.uc3m.es)

Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid
 

Material del seminario:

material-tony.zip

OBJETIVO

En este seminario se analizar�n en profundidad los conceptos m�s importantes de la entrada/salida paralela y su evoluci�n en los �ltimos a�os. Adem�s se estudiar�n las nuevas t�cnicas propuestas para computaci�n en clusters y en rejillas.

Programa:

1.Introduction

2.Sequential file systems

3.Network architectures

4.Data placement

5.Caching & prefetching

6.Server architecture

7.Interface

8.Applications

CV del ponente:

Toni Cort�s es profesor titular del departamento de Arquitectura de Computadores en la Universidad Polit�cnica de Catalu�a. Obtuvo t�tulo de licenciado en inform�tica en 1992 y el de doctor en inform�tica en 1997.

Actualmente, adem�s de sus tareas de investigaci�n y docencia, es vicedecano de relaciones internacionales de la Facultad de Inform�tica de Barcelona y coordinador del area de �single-system Image� de la �IEEE task force� sobre �cluster computing�. El resultado de su tarea investigadora ha producido mas de 25 art�culos en congresos y revistas de prestigio internacional, el libro �High Performance Mass Storage and Parallel I/O: Technologies and Applications� y un cap�tulo sobre temas de entrada/salida en el libro �High Performance Cluster Computing: Architectures and Systems �.

Tambi�n participa regularmente como miembro del comit� de programa de varios congresos de prestigio internacional en los temas de entrada/salida, sistemas operativos, cluster computing y paralelismo en general. Finalmente, tambi�n ha participado en varios proyectos de investigaci�n en cooperaci�n con empresas europeas dentro del marco de proyectos europeos.

 

 

 

Title Speaker Date
POWER AWARE ARCHITECTURES AND PARALLEL I/O SYSTEMS Prof. Alok Choudhary 17 y 18 de marzo de 2003
Abstract
 

Título:POWER AWARE ARCHITECTURES AND PARALLEL I/O SYSTEMS

Prof. Alok Choudhary

Department of Electrical Computing Engineering
Nothwestern University
Chicago, EE.UU.
Fecha impartici�n: 17 y 18 de marzo de 2003.

Horario: de 15 a 20 horas.

Duraci�n: 10 horas.

Aula: 20D14 horas.

Lugar: Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

 

 

Material del seminario:

alok-2003.zip

 

Abstract of the course:

This seminar shos the new trends in high performance computing, including two important new fields: power aware architectures and parallel I/O systems.

Topics:

*Part I. Power Aware Architectures:

1.Power aware Architctures.
2.Power Aware memory Hierarchies.
3.Power Aware Compilation and Optimizations.
4.Synthesis and compilation for Heterogeneous SoC systems
5.Power Aware parallelization for Embedded Systems.

*Part II: Parallel I/O Systems and Trends.

1.Smart-Disk based Storage Archietctures
2.Clusters of Smart-Disk Systems
3.Software structures for Smart Disk based Systems.
4.Parallel File Systems for Clusters
5.Optimizing Parallel I/O for large-scale systems including MPI-IO
6.Scientific Data Management

Short biography of the autor:

Prof. Alok Choudhary got his Ph.D. at the University of Illinois, Urbana-Champaign, in 989. He is currently a full Professor, ECE Department in the Northwestern University, Chicago, USA. He has got several awards, including the NSF Young Investigator Award, in 1993, the IBM Faculty Development Award, in 1994, the Intel Faculty Research Award, in 1993, and the IEEE Engineering Foundation Award, in 1990. His research interest includes compilers and Runtime Systems for High-Performance, embedded and adaptive computing systems and power-aware systems, High-Performance Databases, OLAP and datamining and Parallel and distributed Input-Output for Scientific and Information Processing Applications.

 

 

 

Title Speaker Date
Modelado del comportamiento de sistemas informáticos y de comunicaciones Ramon Puigjaner 11, 12 y 13 de noviembre de 2002
Abstract
 

Título: MODELADO DEL COMPORTAMIENTO DE SISTEMAS INFORM�TICOS Y DE COMUNICACIONES

Ramon Puigjaner

Universidad de las Islas Baleares

Fecha impartici�n: 11, 12 y 13 de noviembre de 2002.

Horario: de 16 a 20 horas. Duraci�n: 10 horas.

Lugar: Aula 20D14 Edificio Sabatini

Responsable: Jes�s Carretero (jcarrete@arcos.inf.uc3m.es)

Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

Material del seminario:

modelado.zip

 

OBJETIVO

Este curso pretende dar una visi�n de que se entiende por modelar el comportamiento de un sistema inform�tico o de comunicaciones. Para ello se revisan variables (datos e inc�gnitas) luego se describe la forma en que pueden construirse modelos que representen el comportamiento din�mico de los sistemas inform�ticos y de comunicaciones.

Program:

1.INTRODUCCI�N

2.NECESIDAD Y CONCEPTO DE LA EVALUACI�N DEL COMPORTAMIENTO DE UN SISTEMA INFORM�TICO

3.VARIABLES

3.1.Del comportamiento

3.2.De la carga

3.3.De control

4.Herramientas: Modelos del comportamiento

4.1.Revisi�n de la teor�a de colas

4.1.1.Procesos de nacimiento-muerte

4.1.2.Colas M/M/n/K/N

4.1.3.Colas M/G/1

4.1.4.Colas G/G/1: Aproximaci�n por difusi�n

4.1.5.Colas con prioridades

4.1.6.Cola de Skinner

4.2.Modelos locales de comportamiento

4.2.1.Tambores

4.2.2.Discos

4.2.3.Subsistemas secuenciales

4.2.4.CPU

4.2.5.Memoria

4.2.6.Sistemas de comunicaciones

4.3.Modelizaci�n global del comportamiento: Teor�a de redes de colas

4.3.1.Teorema de BCMP

4.3.2.M�todos de c�lculo

4.3.3.M�todos aproximados

4.3.3.1.Difusi�n

4.3.3.2.Descomposici�n-agregaci�n

4.3.3.3.Iterativos

4.4.Modelizaci�n global del comportamiento: Simulaci�n

4.4.1.Generaci�n de n�meros aleatorios

4.4.1.1.N�meros aleatorios uniformes

4.4.1.2.N�meros aleatorios seg�n otras distribuciones

4.4.1.3.Tests de validez de las secuencias de n�meros aleatorios

4.4.2.Organizaci�n general del simulador

4.4.3.Determinaci�n de la calidad de los resultados: Intervalos de confianza

4.4.3.1.M�todos de las repeticiones y de las medias de los lotes

4.4.3.2.M�todo regenerativo

4.4.3.3.M�todo espectral

4.5.Estudio de casos

4.5.1.Modelizaci�n del comportamiento de sistemas inform�ticos: Tratamientos anal�ticos y por simulaci�n

4.5.2.Modelizaci�n del comportamiento de sistemas teleinform�ticos: Tratamientos anal�ticos y por simulaci�n

5.Conclusi�n

Título:Curr�culum Vitae del ponente

Catedr�tico de Universidad de Arquitectura y Tecnolog�a de Computadores del Departamento de Ciencias Matem�ticas e Inform�tica de la Universidad de las Islas Baleares y Director de la Escuela Polit�cnica Superior. Es Doctor Ingeniero Industrial y Licenciado en Inform�tica.

Su inter�s investigador actual se centra en la evaluaci�n del comportamiento de sistemas inform�ticos y de comunicaciones y la difusi�n de estas t�cnicas en el �mbito industrial, principalmente en el �mbito de los sistemas de tiempo real, de los sistemas concurrentes y de los sistemas distribuidos.

Es autor de un libro sobre evaluaci�n de prestaciones y de m�s de un centenar trabajos entre cap�tulos de libros, art�culos en revistas y presentaciones (evaluadas y publicadas) en congresos.

 

 

 

Title Speaker Date
Quality of Service (QoS) in the Internet: Technologies and Trends Dr. Hermann HELLWAGNER 20 – 22 de febrero de 2002
Abstract
 

Título:Quality of Service (QoS) in the Internet: Technologies and Trends

Univ.-Prof. DI Dr. Hermann HELLWAGNER
Dept. of Information Technology
University Klagenfurt
Klagenfurt, Austria

Fecha impartici�n: 20 – 22 Febrero de 2002
20 Feb: 4 horas; 21 Feb: 3 horas; 22 Feb: 3 horas.
Horario: de 16 a 20 horas.

Duraci�n: 10 horas.

lugar: Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

AULA: 20D10. Edificio Sabatini.

 

 

Material del seminario:

qos.zip

 

Short biography of the author:

Hermann Hellwagner received his Dipl.-Ing. degree (in Informatics) and Ph.D. degree (Dr. techn.) in 1983 and 1988, respectively, both from the University Linz, Austria.

From 1989 to 1994, he was senior researcher and team/project manager at Siemens AG, Corporate Research and Development, M�nchen, Germany, working on distributed shared memory, parallel file systems, and performance analysis of computer and communication systems. From 1995 to 1998, he was associate professor of parallel computer architecture at Technische Universit�t M�nchen (TUM) where he led a group of researchers working on compute clusters and lightweight protocols and communication libraries based on the high-performance Scalable Coherent Interface (SCI) system area network. Since late 1998, Dr. Hellwagner has been a full professor of computer science in the Department of Information Technology at the University Klagenfurt, Austria. His current areas of interest are multimedia communication, cluster computing, embedded systems, and hardware/software interaction. Current projects are on digital video communication and adaptation and a high-end thin client specifically targeted at CAD applications.

Dr. Hellwagner is co-editor of a book (on SCI) in the Springer LNCS State-of-the-Art Survey series, (co-)editor of five workshop proceedings, author of seven book chapters, and (co-)author of about 40 conference and journal papers and of several technical reports and project reports. He is Subject Area Editor for Computer Architecture of the Journal of Systems Architecture (JSA). He initiated two annual international workshop series (HIPS and SCI-Europe), and has served as a program committee member for several international conferences (recently, for FTPDS’2001, EBP�01, Cluster’2001, NCA�01, and CCGrid’2001) and a reviewer for several journals and many conferences.

Dr. Hellwagner is a member of IEEE, GI (German Informatics Society), and OCG (Austrian Computer Society). He is also the head of the Austrian delegation to the ISO/IEC MPEG standardization group.

Abstract of the course:

Quality of Service (QoS) mechanisms become increasingly important in computer and communication networks, both for the transmission of multimedia data (e.g., audios and videos, with real-time constraints) and for business-critical data (e.g., transactions in e-commerce systems, with high-reliability requirements). In recent years, intense research and standardization activities have been carried out to establish QoS mechanisms and technologies for the (future) Internet which, by way of its original design and evolutionary growth, provides best-effort data transfers only. This seminar will give a motivation for Internet QoS, present the principal technologies and protocols devised and standardized by the IETF to provide QoS guarantees, and discuss the current status and potential trends of QoS mechanisms in IP-based networks. A case study will also be discussed.

Topics:

1. Motivation for Internet Qos

2. QoS Principles and Terminology (definitions, requirements, parameters,
specification, negotiation, resource reservation and management, traffic
shaping and policing,�)

3. QoS Architectures

4. Queueing Techniques for QoS

5. Integrated Services (IntServ), Resource Reservation Protocol (RSVP)

6. Differentiated Services (DiffServ)

7. Multiprotocol Label Switching (MPLS)

8. Protocols for Real-Time Data Transfers (RTP, RTCP, RTSP)

9. Case Study: Internet Telephony

10. Discussion: Status and Potential Trends

Prerequisites of students:

The students are expected to have a working knowledge of IP-based networks and the TCP/IP protocol family.

References:

John Barnes, Programming in Ada 95, 2nd. Ed, Addison-Wesley, 1998.

 

 

 

Title Speaker Date
High-performance OLAP and Data Mining Systems, Techniques and Applications Prof. Alok Choudhary 20 y 21 de marzo de 2002
Abstract
 

Título:High-performance OLAP and Data Mining Systems,Techniques and Applications

Prof. Alok Choudhary

Department of Electrical Computing Engineering
Nothwestern University
Chicago, EE.UU.
Fecha impartici�n: 20 y 21 de marzo de 2002.

Horario: de 16 a 21 horas.

Lugar: Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

Edificio Torres Quevedo. Aula: 40F16.

Material del seminario:

alok.zip

Short biography of the autor:

Prof. Alok Choudhary got his Ph.D. at the University of Illinois, Urbana-Champaign, in 989. He is currently a full Professor, ECE Department in the Northwestern University, Chicago, USA. He has got several awards, including the NSF Young Investigator Award, in 1993, the IBM Faculty Development Award, in 1994, the Intel Faculty Research Award, in 1993, and the IEEE Engineering Foundation Award, in 1990. His research interest includes compilers and Runtime Systems for High-Performance, embedded and adaptive computing systems and power-aware systems, High-Performance Databases, OLAP and datamining and Parallel and distributed Input-Output for Scientific and Information Processing Applications.

Abstract of the course:

OLAP and data mining are integral parts of any decision support process. Yet, to date, most OLAP systems have focused on providing �access� to multi-dimensional data, while data mining systems have dealt with �influence analysis� of data along a single dimension. In this article, I show that OLAP and data mining should not remain separate components of decision support, but should be fully merged � they are inherently related activities that re-inforce (and desperately need) each other. This course focuses on both topics and their relations, specifically to deal with Large-Scientific datasets.

Topics:

1.Data Mining

2.On-Line-Analytical Processing

3.Data Warehousing (basic Design and Design for E-commerce Click stream)

4.Parallel OLAP

5.Parallel Data Mining

6.High-Performance Datawarehousing

7.Applications of the above technologies in various domain

8.New Approaches to dealing with Large-Scientific datasets

9.Large-scale Scientific data analysis

10.Applications

 

 

 

Title Speaker Date
Router para tráfico multimedia Prof. Francisco José Quiles 7 y 8 de marzo de Marzo de 2002
Abstract
 

T�tulo:Router para tr�fico multimedia

Prof. Francisco Jos� Quiles

Departmento de Inform�tica

Universidad de Castilla-La Mancha

Albacete. Espa�a.
Fecha impartici�n: 7 y 8 de marzo de Marzo de 2002.

Horario: de 15:30 a 20 horas.

Duraci�n: 10 horas.
Lugar: Escuela Polit�cnica Superior. Aula: 20D10

Universidad Carlos III.

Legan�s, Madrid
Material del seminario:

quiles.zip

 

Breve biograf�a del autor:

El Prof. Francisco Jos� Quiles Flor es Catedr�tico de Escuela Universitaria en el �rea de Conocimiento: Arquitectura y Tecnolog�a de Computadores. Imparte docencia e investiga en temas realacionados con la Arquitectura e Ingenier�a de Computadores, Arquitecturas no Cl�sicas y Redes de Altas Prestaciones. Su principal tema de inter�s en investigaci�n es el de encaminadores para tr�fico multimedia. Actualmente ocupa el cargo de Vicerrector de Investigaci�n en la Universidad de Castilla La Mancha.

Resumen del curso:

En este seminario se realiza un estudio del estado del arte en el dise�o de encaminadores orientados al tr�fico multimedia, repasando uno a uno todos los aspectos que se deben tener en cuenta para el dise�o de estos encaminadores, teniendo en cuenta las restricciones que impone este tipo de tr�fico. Centr�ndonos en las caracter�sticas m�s apropiadas para entornos de �rea local. Se analizan no s�lo la evaluaci�n de prestaciones de cada una de las soluciones, sino tambi�n la incidencia que un determinado aspecto puede tener en la complejidad del circuito resultante. Finalizando con un ejemplo pr�ctico de arquitectura orientada al tr�fico multimedia y analizando las consecuencias de la nueva arquitectura InfiniBand.

Abstract:

1.Estado del arte

2.Dise�o de encaminadores para multimedia

3.Soluciones existentes

4.Evaluaci�n de prestaciones

5.Complejidad de los circuitos

6.Arquitecturas para tr�fico multimedia

7.Ejemplo pr�ctico

8.Incidencia de InfiniBand

 

 

 

Title Speaker Date
Tool Sharing in Multimedia Collaborative Applications Prof. Hussein Abdel-Wahab 22-24 de Abril de 2002
Abstract
 

T�tulo:Tool Sharing in Multimedia Collaborative Applications

Prof. Hussein Abdel-Wahab

Department of Computer Science
Old Dominion University
Norfolk, Virginia, EE.UU.
Fecha impartici�n: 22-24 de Abril de 2002.

Horario: de 16 a 20 horas.

Duraci�n: 10 horas.

Lugar: Escuela Polit�cnica Superior

Edificio Sabatini. Aula 20D10
Universidad Carlos III.

Legan�s, Madrid

Material del seminario:

madridseminar.zip

Short Biography of the speaker:

Hussein Abdel-Wahab received the Ph.D in 1976 and the MS in 1973 both from the University of Waterloo in Computer Communications and the BS in Electrical Engineering from Cairo University in 1969. Currently he is a full professor of computer science at Old Dominion University. In addition he is an adjunct professor of computer science at the University of North Carolina at Chapel Hill and a faculty at the Information Technology Lab of the National Institute of Standards and Technology. Prior to that he held faculty positions at North Carolina State University, the University of Maryland and Rochester Institute of Technology. He served as a consultant to many organizations including IBM, MCNC and MITRE Corp. He is the principle investigator in designing and implementation of XTV, a pioneer X-window based Teleconferencing system. His main research interests are collaborative desktop multimedia conferencing systems, and real-time distributed information sharing. His research has been supported by NSF, ONR, IBM, MCNC, MITRE, ARPA among others. He is a senior member of IEEE Computer Society and a member of the Association for Computing Machinery.

Abstract of the course:

Increased desktop processing power and network bandwidth have made feasible distributed multimedia collaborative systems. Such systems are characterized by video, audio and data exchanges. While there has been much research and development of protocols and services for video and audio transmission, less work has been focused on data sharing particularly when the data is being generated by an application in real time and in turns controlled by one or more of the participants in the collaborative session. This series of seminars describes the principles, protocols and implementation for sharing desktop applications in a distributed collaborative session. Examples of such systems are X Teleconferencing and Viewing (XTV) and Java Collaborative Environment (JCE) which have been implemented at Old Dominion University will be described in these series of seminars.

Topics:

1. Introduction to sharing desktop applications

2. Principles of sharing desktop applications

3. Protocols for sharing desktop applications in a distributed collaborative session

4. Implementation of sharing desktop applications in a distributed collaborative session

5. Examples: X teleconferencing and Viewing (XTV) and Java Collaborative Environment (JCE)

 

 

 

Title Speaker Date
Sistemas de Tiempo Real Juan Antonio de la Puente y Alejandro Alonso 17-19 de Enero de 2001
Abstract
 

Título:SISTEMAS DE TIEMPO REAL

Docentes: Juan Antonio de la Puente y Alejandro Alonso

Universidad Polit�cnica de Madrid

Fecha impartici�n: 17-19 de Enero de 2001.

Horario: de 16 a 20 horas (D�as 17 y 18). 16 a 18 horas (D�a 19).

Duraci�n: 10 horas.

Lugar: Aula 20D14 Edificio Sabatini

Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

Material del seminario:

PROGRAMA y MATERIAL DEL SEMINARIO

Objetivos:

Los sistemas inform�ticos de tiempo real se utilizan en numerosos campos de aplicaci�n, tales como el control de aviones, autom�viles y trenes, tr�fico, comunicaciones, sat�lites, control de procesos, electr�nica de consumo, etc. La caracter�stica m�s importante de estos sistemas es que sus acciones se deben ejecutar en intervalos de tiempo determinados por la din�mica de los sistemas f�sicos que supervisan o controlan. Adem�s, suelen tener requisitos estrictos de fiabilidad y seguridad.

Muchos de estos sistemas est�n empotrados en otros sistemas, lo que conlleva una limitaci�n de recursos (potencia de procesador, memoria, etc.) con respecto a otros tipos de sistemas inform�ticos. En la mayor�a de los casos, adem�s, deben ejecutar varias tareas concurrentemente. Estas caracter�sticas, en conjunto, hacen que la construcci�n de sistemas de tiempo real presente caracter�sticas espec�ficas respecto a otros tipos de sistemas inform�ticos.

El curso trata adem�s aspectos de sistemas distribuidos de tiempo real.

Título:Temario

1.Introducci�n a los Sistemas de Tiempo Real

2.Programaci�n de Sistemas de Tiempo Real: Introducci�n a C y Ada

3.Sistemas C�clicos

4.Sistemas Multiprogramados

5.Planificaci�n de Tareas

6.Comunicaci�n y Sincronizaci�n

7.Programaci�n de bajo nivel

8.Sistemas Distribuidos

9.Dise�o de Sistemas de Tiempo Real

10.Aplicaciones

Bibliograf�a:

Alan Burns & Andy Wellings, Real-Time Systems and Programming Languages, 2nd ed., Addison-Wesley, 1996. ISBN 0-201-40365-X

Hermann Kopetz, Real-Time Systems. Design Principles for Distributed Embedded Applications, Kluwer, 1997. ISBN 0-7923-9894-7.

John Barnes, Programming in Ada 95, 2nd. Ed, Addison-Wesley, 1998.

 

 

 

Title Speaker Date
EVALUACIÓN DEL COMPORTAMIENTO DE SISTEMAS INFORMÁTICOS Y DE COMUNICACIONES Ramon Puigjaner 4 y 5 de Diciembre de 2000
Abstract
 

T�tulo: EVALUACI�N DEL COMPORTAMIENTO DE SISTEMAS INFORM�TICOS Y DE COMUNICACIONES

Ramon Puigjaner

Universidad de las Islas Baleares
Fecha impartici�n: 4 y 5 de Diciembre de 2000.

Horario: de 15 a 20 horas. Duraci�n: 10 horas.

Lugar: Aula 20D14 Edificio Sabatini

 

Escuela Polit�cnica Superior

Universidad Carlos III.

Legan�s, Madrid

 

Material del seminario:

evaluacion.html

 

OBJETIVO

Este curso pretende dar una visi�n de lo que se entiende por evaluar el comportamiento de un sistema inform�tico o de comunicaciones. Para ello se revisan variables (datos e inc�gnitas) que hay que manipular y las distintas herramientas (monitores, benchmarks y modelos) que pueden utilizarse para lograr el fin deseado seg�n cual sea el objetivo del estudio y el entorno de trabajo del sistema.

Título:Programa

1.INTRODUCCI�N

2.NECESIDAD Y CONCEPTO DE LA EVALUACI�N DEL COMPORTAMIENTO DE UN SISTEMA INFORM�TICO

3. ENTORNOS DE EVALUACI�N DEL COMPORTAMIENTO

3.1. Seg�n la forma de trabajo

3.2. Seg�n el objetivo perseguido

4. VARIABLES

4.1. Del comportamiento

4.2. De la carga

4.3. De control

5. HERRAMIENTAS: MONITORES

5.1. Formas de medici�n

5.2. Caracter�sticas de los monitores

5.3. Tipos de monitores

5.4. Presentaci�n de resultados

5.5. Monitores de redes locales

6. Herramientas: Modelos de la carga (benchmark)

6.1. Representatividad de los modelos de la carga

6.2. Formas de los modelos de la carga

6.3. Construcci�n de modelos de la carga

6.4. Realizaci�n pr�ctica de benchmarks

6.5. Revisi�n de los benchmarks est�ndar

7. MODELADO: aN�LISIS OPERACIONAL

7.1. Variables y leyes operacionales

7.2. An�lisis de cuellos de botella

7.3. Ejemplos

8. CONCLUSI�N: REVISI�N DEL PUNTO 3 Y DISCUSI�N

 

 

 

Title Speaker Date
Sistemas Tolerantes a Fallos (STF) Pedro Gil y Juan José Serrano 5 de Junio de 2000.
Abstract
 

Título:Sistemas Tolerantes a Fallos (STF)

Docentes: Pedro Gil y Juan Jos� Serrano

Universidad Polit�cnica de Valencia

Fecha impartici�n: 5 de Junio de 2000.

Duraci�n: 10 horas.

MATERIAL DE SEMINARIO.

Objetivos:

* Conocer los conceptos y terminolog�a b�sicos referentes a la garant�a de funcionamiento[1] o confiabilidad ( fiabilidad, seguridad-inocuidad, seguridad-confidencialidad, disponibilidad, mantenibi-lidad, etc.) de los sistemas inform�ticos.

* Conocer y comprender las t�cnicas de redundancia (mecanismos de tolerancia a fallos), tanto a nivel f�sico como l�gico, que permiten dise�ar sistemas de funcionamiento garantizado.

* Conocer las arquitecturas b�sicas de los STF, en sistemas monoprocesadores, multiprocesadores y distribuidos.

* Estudiar las t�cnicas de modelado y evaluaci�n de la garant�a de funcionamiento de los sistemas tolerantes a fallos.

* Conocer ejemplos de sistemas tolerantes a fallos comerciales, as� como sus principales campos de aplicaci�n: Sat�lites espaciales, avi�nica, medicina, comunicaciones, banca, comercio electr�nico

Título:Temario

Tema 1 : Introducci�n a la garant�a de funcionamiento. Definiciones fundamentales:

1.1.- Definici�n de la garant�a de funcionamiento.

1.2.- Necesidad de los sistemas con alta garant�a de funcionamiento. Atributos de la garant�a de funcionamiento.

* Fiabilidad , Seguridad � inocuidad, Seguridad � confidencialidad, Disponibilidad, Mantenibilidad.

1.3.- �rbol de la garant�a de funcionamiento.

1.4.- Causas de la disminuci�n de la garant�a de funcionamiento.

* Fallos.

* Errores.

* Aver�as.

* Patolog�a de los fallos

1.5.- Medios para obtener cierta garant�a de funcionamiento:

* Prevenci�n de fallos

* Tolerancia a fallos.

1.7.- Medios para la validaci�n de la garant�a de funcionamiento

* Eliminaci�n de fallos.

* Predicci�n de fallos.

Tema 2.- T�cnicas de redundancia:

2.1. T�cnicas de redundancia en la informaci�n:

* Introducci�n a los c�digos detectores y correctores de errores.

* C�digos de paridad.

* Sumas de prueba.

* C�digos m de n.

* C�digos c�clicos. CRCs

* C�digos aritm�ticos.

* C�digos correctores de errores de Hamming.

2.2. T�cnicas de redundancia en el hardware:

* Introducci�n.

* Redundancia pasiva. Principios b�sicos.

* Redundancia activa. Principios b�sicos.

* Redundancia h�brida.

2.3. T�cnicas de redundancia en el software:

* Introducci�n.

* Redundancia temporal

* Redundancia en el software para detecci�n de fallos aleatorios

* Mecanismos de tolerancia a fallos de dise�o en el software.

* Tolerancia a fallos en el Sistema Operativo.

Tema 3.- Arquitectura de los sistemas tolerantes a fallos:

3.1.- Sistemas monoprocesadores tolerantes a fallos:

* Detecci�n de errores por el propio microprocesador.

* Detecci�n de errores por control del flujo del programa.

* Recuperaci�n de los errores en sistemas monoprocesador.

3.2.- Sistemas multiprocesadores tolerantes a fallos.

* Introducci�n.

* Sistemas con redundancia est�tica.

* Sistemas con redundancia din�mica o �standby�.

* Sistemas de detecci�n de errores no est�ndares.

* Recuperaci�n de los errores sistemas multiprocesador.

3.3.- Sistemas distribuidos tolerantes a fallos.

* Introducci�n.

* Comunicaciones fiables.

* Procesadores con parada tras la aver�a

* Arquitecturas de sistemas distribuidos tolerantes a fallos.

* Sistemas de detecci�n de errores no est�ndares.

* Recuperaci�n de los errores sistemas distribuidos.

Tema 4.- Modelado y evaluaci�n de los sistemas tolerantes a fallos:

4.1 Introducci�n.

4.2 Funciones para la evaluaci�n de STFs.

4.3 C�lculo de la tasa de fallos y de la tasa de reparaci�n.

4.4 T�cnicas de modelado.

* Modelos combinatorios.

* Cadenas de Markov.

4.5 Modelado con coberturas.

4.6 Modelos de Markov.

* Modelos para la fiabilidad.

* Modelos para la seguridad.

* Modelos para la disponibilidad.

4.7 Ejemplos de modelado y evaluaci�n te�rica.

4.8 Introducci�n a la validaci�n experimental de STFs: Inyecci�n de fallos.

Tema 5.- Aplicaciones y ejemplos de sistemas tolerantes a fallos:

5.1 Sistemas de larga vida. Sat�lites espaciales.

* STAR (Self Testing And Repairing Computer)

* Sistema de la sonda espacial Voyager

* Sistema de la sonda espacial Galileo

5.2 Sistemas cr�ticos. Avi�nica, industrias peligrosas.

* Sistema del Space Shuttle

* Sistema flying by wire de los Airbus

5.3 Sistemas de alta disponibilidad.

* Tandem NonStop Cyclone

* Stratus XA/R serie 300

* Sequoia Serie 400

Bibliograf�a:

[Arlat 96] J. Arlat y otros, coordinados por J. C. Laprie:�Guide de la s�ret� de fonctionement�. Ed. C�padu�s editions, Toulouse, Francia, 1996.

[Johnson 89] B. W. Johnson: Design and analysis of fault tolerant digital systems. Ed. Addison�Wesley, 1989

[Gil96] P. J. Gil: Garant�a de funcionamiento, conceptos b�sicos y terminolog�a. Informe interno, DISCA�GSTF, 1996

[Laprie92] J. C. Laprie: Dependable computing: Basic concepts and terminology: in english, french, german, italian and japanese. Ed. Springer�Verlag, 1992

[Pradhan86] D. K. Pradhan: Fault tolerant computing: Theory and techniques; vol. 1 y 2. Ed. Prentice Hall, 1986

[Pradhan96] D. K. Pradhan: Fault-Tolerant Computer System Design. Ed. Prentice Hall, 1996

[Siewiorek92] D.P. Siewiorek, R. S. Schwarz: Reliable computer systems: Design and evaluation (2nd. Edition). Ed. Digital Press, 1992

[1] En ingl�s �Dependability�.