INSTITUTO DE COMPUTAÇÃO
. |
Computação Aproximada com Disciplina para Eficiência de Energia: da Linguagem de Programação ao Hardware.Local:
Auditório do IC - Sala 85 - IC 2
Palestrante: Palestra: On ESL Verification of Memory Consistency for System-on-Chip Multiprocessing.Local:
Auditório do IC - Sala 85 - IC 2
Palestrante: Prof. Dr. Luiz C. V. dos Santos (UFSC) Chip multiprocessing is key to Mobile and highend Embedded Computing. It Prof. Dr. Luiz C. V. dos Santos (UFSC): ====================================================== Palestra Série Seminários 2012: SMOOTH SIGNED DISTANCE SURFACE RECONSTRUCTION AND APPLICATIONS - 17/08/2012Local:
Auditório IC 2 - Sala 85
SMOOTH SIGNED DISTANCE SURFACE RECONSTRUCTION AND APPLICATIONS Prof. Gabriel Taubin Division of Engineering Brown University Providence, RI, USA RESUMO In this talk I will describe a new and simple variational formulation for the problem of reconstructing the surface geometry, topology, and color map of a 3D scene from a finite set of colored oriented points. These data sets are nowadays obtained using a variety of techniques, including 3D shape capture systems based on structured lighting, pasive multi-view stereo algorithms, and 3D laser scanning. In this formulation the implicit function is forced to be a smooth approximation of the signed distance function to the surface. The formulation allows for a number of different efficient discretizations, reduces to a finite dimensional least squares problem for all linearly parameterized families of functions, and does not require the specification of boundary conditions. The resulting algorithms are significantly simpler and easier to implement than alternative methods. This method is particularly good at extrapolating missing and/or irregularly sampled data. An efficient implementation based on a primal-graph octree-based hybrid finite element-finite difference discretization, and the Dual Marching Cubes isosurface extraction algorithm, is shown to produce high quality crack-free adaptive manifold polygon meshes. After the geometry and topology have been reconstructed, the method then smoothly extrapolates the color information from the points to the surface. Experimental evidence is presented to show that the resulting method produces high quality polygon meshes with smooth color maps, which accurately approximate the source colored oriented points. An open source implementation of this method is available for download. I will conclude describing applications to digital archaeology and 3D forensics. Organizador: Prof. Siome Goldenstein IC -- Unicamp Fone: (019) 3521-5888 Palestra Extraordinária: DYNAMIC TEMPORAL WORKLOAD in HYBRID DATA CENTERS and ENERGY-AWARE AGGREGATIONLocal:
Sala 85 - IC
Prof. Deep Medhi,
Computer Science & Electrical Engineering Department,
University of Missouri-Kansas City, USA
(e-mail:DMedhi@umkc.edu)
RESUMO
Workloads in data centers are highly dynamic. Secondly, data centers are often hybrid in terms of hardware/compute capability. An option at one extreme is to keep all machines running all the time – the downside is that cost of running (including energy cost) becomes very high. On the other hand, if workload is broken down into review points (say every 5 minutes), then based on the aload some servers may be put in sleep mode (to reduce energy cost); but this then bring up an additional cost facture due to machine turn-on/off, which for example reduces the hardware lifetime. Thus, it is important to understand how to optimally use resources in a data center so that the cost can be minimized over a time window. It may be noted that it may be possible to merge review points to extend the span of review points, if the workload doesn’t change drastically, or we decide to aggregate them to potentially reduce cost. However, this is not as simple. Due to aggregation, the energy cost may actually go up.
Thus, we have considered the problem of balancing energy consumption and system cost in hybrid data centers with dynamic temporal workloads. For comparison, we consider three data-center scenarios: all homogeneous (i.e., all machines are of the same type), all heterogeneous (each one different), mixed clusters different clusters, where each cluster has homogenous nodes). Specifically, for each data center type, we’ll present time dependent optimization models that capture the workload requirements as well as the different cost factors.
Through our study, we found that the computational time for the heterogeneous model is most time consuming. Thus, to cut this down, some aggregation is necessary; on the other hand, as point out earlier, this increases the energy cost. Thus, we will discuss the trade-off between energy-aware aggression and the computational cost, and where and how a dynamic aggregation scheme helps compared to a static aggregation scheme.
Bio: Deep Medhi is Curators' Professor in the Department of Computer Science and Electrical Engineering at the University of Missouri- Kansas City, USA. He received B.Sc. in Mathematics from Cotton College, Gauhati University, India, M.Sc. in Mathematics from the University of Delhi, India, and his Ph.D. in Computer Sciences from the University of Wisconsin-Madison, USA. Prior to joining UMKC in 1989, he was a member of the technical staff at AT&T Bell Laboratories. He was an invited visiting professor at the Technical University of Denmark, a visiting research fellow at Lund Institute of Technology, Sweden, and a Fulbright Senior Specialist. He is the Editor-in-Chief of Springer’s Journal of Network and Systems Management, and is on the editorial board of IEEE/ACM Transactions on Networking, IEEE Transactions on Network and Service Management, and IEEE Communications Surveys & Tutorials. He has published over a hundred papers, and is co-author of the books, Routing, Flow, and Capacity Design in Communication and Computer Networks (2004) and Network Routing: Algorithms, Protocols, and Architectures (2007), both published by Morgan Kaufmann Publishers.
Palestra Extraordinária: Quality-aware Dynamic Software Product Lines; the Body Area Network CaseLocal:
Sala 85 - IC
Prof. Vander Alves
Departamento de Ciência da Computação
Universidade de Brasilia
RESUMO
Demographic and social changes have increased the number of elderly people living alone. Many of these need continuous medical assistance, yet it is not sustainable to have a dedicated medical professional for each of them. Accordingly, automated support has been proposed, in particular, Body Area Network, in which a person goes about her daily activities at home or outdoors, but wears sensors monitoring her vital signs and providing emergency detection and prevention. Such systems have to strike a balance between conflicting requirements, such as availability and reliability, e.g, if a person is well, not all sensors need to be active and working at the highest sampling rate; on the other hand, if she falls or has a stroke, the opposite must happen with the sensors. We explore how Dynamic Software Product Line (DSPL) achieve this goal. A DSPL reconfigures itself based on some context change e.g., the persons' medical situation, to meet a new quality goal for that new situation, as specified by a reliability contract provided by the domain expert (a medical doctor). This contract is modeled as a state machine, whose transitions are medical events (e.g., fall, stroke) and states are target reliability goals, prompting a reconfiguration to meet it. The quality of any given configuration is measured by a single formula, parametrizing over the features of the DSPL and related quality information. This formula is derived from a parametric discrete time Markov chain model representing the reliability of the DSPL. Further, this formula relies on a normalized form of the feature model. We analyse achieved expressiveness as well as time and space complexity issues.
Short Bio:
Vander Alves é Professor Adjunto do Departamento de Ciência da Computação da Universidade de Brasília. Ele se interessa por pesquisa, desenvolvimento, e ensino em Engenharia de Software, em particular nas sub-áreas: Linha de Produtos de Software, Ambient Assited Living, e Desenvolvimento de Software Orientado a Aspectos e Modelos. Vander é doutor em Ciência da Computação (Engenharia de Software) pela Universidade Federal de Pernambuco. Ele foi pesquisador pós-doutor no Fraunhofer Institute for Experimental Software Engineering (Alemanha) e na Lancaster University (Inglaterra), tendo trabalhado em projetos europeus nas áreas de Linha de Produtos de Software, Desenvolvimento de Software Orientado a Aspectos e a Modelos, Ambient Assited Living, e Smart Homes. Ele também trabalhou no IBM Silicon Valley laboratory em San Jose, Califórnia (EUA), na implementação da linha de produtos Information Integration, e é co-autor de uma patente licenciada nos EUA.
Organizadora: Profa. Cecilia Rubira ( cmrubira [at] ic [dot] unicamp [dot] br )
IC -- Unicamp Fone: (019) 3521-5879
Palestra Extraordinária: Protection for Three-Layer IP/MPLS Over OTN Over DWDM Networks.Local:
Auditório do IC - Sala 85 - IC 2
Série de Seminários da Pós-Graduação do IC - 2012 Palestrante: Prof. Deep Medhi, Computer Science & Electrical Engineering Department,
University of Missouri-Kansas City, USA
==================================================
For a number of years, multilayer networking has been a topic of great interest among researchers due to do the inherent nature of
multilayer topological structure in deployed networks. For example, there have been many works on IP over DWDM to show how IP raffic is carried over DWDM networks. Such works usually consider the design and interaction between two adjacent layers, namely P and DWDM. With the deployment of MPLS, there has been also work on IP/MPLS over DWDM.
A recent entrant to multilayer networking is OTN (optical transport network) technology. What OTN allows is to provide a digital wrapper functionality between IP and DWDM (which runs in analog mode), thus creating a three-layer network with IP/MPLS over TN over DWDM. Thus, with the introduction of OTN and its various modular capability at different rates, it becomes necessary to understand how a network can be designed when the multi-layer architecture is formed of three layers: IP/MPLS, OTN, and DWDM. Furthermore, since OTN allows the possibility to have sublayer signals that can be used for aggregation, there is a need to develop odels that accurately capture such features. However, there has been very limited work on understanding how to design three layer etworks such as IP/MPLS over OTN over DWDM.
With such multilayering, it is very important to understand how a failure may be propagated among the layers. In addition, a failure at the DWDM layer may percolate through both OTN and IP/MPLS layers because of dependency caused by shared risk link groups SRLG). Thus, it is important to develop network protection model that can capture theses issues.
Our work addresses a protection mechanism at each layer while considering the technological constraints of each layer to develop an ntegrated protection design model, especially considering the OTN layer, which has not been explicitly considered before. We have developed an optimization model and a heuristic approach to observations based on varying several network parameters to nderstand their impacts on protection capacity and the overall network cost. In particular, we will discuss the situations when rotection is not incorporated in every layer.
==================================================
Deep Medhi is Curators' Professor in the Department of Computer Science and Electrical Engineering at the University of Missouri- ansas City, USA. He received B.Sc. in Mathematics from Cotton College, Gauhati University, India, M.Sc. in Mathematics from the niversity of Delhi, India, and his Ph.D. in Computer Sciences from the University of Wisconsin-Madison, USA. Prior to joining UMKC in 989, he was a member of the technical staff at AT&T Bell Laboratories. He was an invited visiting professor at the Technical niversity f Denmark, a visiting research fellow at Lund Institute of Technology, Sweden, and a Fulbright Senior Specialist. He is the Editor-in- hief of Springer’s Journal of Network and Systems Management, and is on the editorial board of IEEE/ACM Transactions on etworking, IEEE Transactions on Network and Service Management, and IEEE Communications Surveys & Tutorials. He has ublished over a hundred papers, and is co-author of the books, Routing, Flow, and Capacity Design in Communication and omputer etworks (2004) and Network Routing: Algorithms, Protocols, and Architectures (2007), both published by Morgan Kaufmann ublishers.
==================================================
Responsável: Prof. Dr. Nelson Luis Saldanha da Fonseca
IC / Unicamp
E-mail: nfonseca@ic.unicamp.br
Fone: (019) 3521-5878
==================================================
|
|
 Webmaster |
| Instituto de Computação :: Universidade Estadual de Campinas :: Av. Albert Einstein, 1251 - Cidade Universitária, Campinas/SP - Brasil, CEP 13083-852 • Fone: [19] 3521-5838 |