http://hdl.handle.net/2117/3125 Fri, 24 May 2013 11:02:13 GMT 2013-05-24T11:02:13Z webmaster.bupc@upc.edu Universitat Politècnica de Catalunya. Servei de Biblioteques i Documentació no http://hdl.handle.net/2117/17911 Title: Operación stencil en CUDA Authors: Garcés, Bernardo; Herrero Zaragoza, José Ramón; Otero Calviño, Beatriz Abstract: Los problemas derivados de la disipación de energía en la computación secuencial, están haciendo que cada vez se popularice más el uso de máquinas y sistemas con mayor cantidad de núcleos de proceso, pasando desde pequeños procesadores con un número reducido de núcleos, por clusters con varias máquinas secuenciales distribuidas, e incluso por dispositivos de coprocesamiento gráfico con varios cientos de núcleos que permiten asignar tareas generales a estos. Muchos algoritmos están siendo adaptados a estos modelos de paralelización. Thu, 21 Feb 2013 12:19:45 GMT http://hdl.handle.net/2117/17911 2013-02-21T12:19:45Z Garcés, Bernardo; Herrero Zaragoza, José Ramón; Otero Calviño, Beatriz no Los problemas derivados de la disipación de energía en la computación secuencial, están haciendo que cada vez se popularice más el uso de máquinas y sistemas con mayor cantidad de núcleos de proceso, pasando desde pequeños procesadores con un número reducido de núcleos, por clusters con varias máquinas secuenciales distribuidas, e incluso por dispositivos de coprocesamiento gráfico con varios cientos de núcleos que permiten asignar tareas generales a estos. Muchos algoritmos están siendo adaptados a estos modelos de paralelización. http://hdl.handle.net/2117/16308 Title: Vectorized register tiling Authors: Berna Juan, Alejandro; Jiménez Castells, Marta; Llaberia Griñó, José M. Abstract: In the last years, there has been much effort in commercial compilers (icc, gcc) to exploit efficiently the SIMD capabilities and the memory hierarchy that the current processors offer. However, the small numbers of compilers that can automatically exploit these characteristics achieve in most cases unsatisfactory results. Therefore, the programmers often need to apply by hand the optimizations to the source code, write manually the code in assembly or use compiler built-in functions (such intrinsics) to achieve high performance. In this work, we present source-to-source transformations that help commercial compilers exploiting the memory hierarchy and generating efficient SIMD code. Results obtained on our experiments show that our solutions achieve as excellent performance as hand-optimized vendor-supplied numerical libraries (written in assembly). Fri, 20 Jul 2012 15:21:50 GMT http://hdl.handle.net/2117/16308 2012-07-20T15:21:50Z Berna Juan, Alejandro; Jiménez Castells, Marta; Llaberia Griñó, José M. no In the last years, there has been much effort in commercial compilers (icc, gcc) to exploit efficiently the SIMD capabilities and the memory hierarchy that the current processors offer. However, the small numbers of compilers that can automatically exploit these characteristics achieve in most cases unsatisfactory results. Therefore, the programmers often need to apply by hand the optimizations to the source code, write manually the code in assembly or use compiler built-in functions (such intrinsics) to achieve high performance. In this work, we present source-to-source transformations that help commercial compilers exploiting the memory hierarchy and generating efficient SIMD code. Results obtained on our experiments show that our solutions achieve as excellent performance as hand-optimized vendor-supplied numerical libraries (written in assembly). http://hdl.handle.net/2117/16307 Title: Source code transformations for efficient SIMD code generation Authors: Berna Juan, Alejandro; Jiménez Castells, Marta; Llaberia Griñó, José M. Abstract: Despite the effort inverted the last years in commercial compilers to generate efficient SIMD instructions based code sequences from conventional sequential programs, the small numbers of compilers that can automatically use these instructions achieve in most cases unsatisfactory results. This work shows how exposing register level reuse in source codes helps vectorizing compilers as ICC to generate efficient SIMD code that exploits vectorial register reuse. To this end we use reverse-engineering to know the limitations of ICC compiler. We compare performance of compiler generated code to performance of hand-optimized assembly-written numerical libraries achieving around the 72% performance of MKL for example. Fri, 20 Jul 2012 13:54:40 GMT http://hdl.handle.net/2117/16307 2012-07-20T13:54:40Z Berna Juan, Alejandro; Jiménez Castells, Marta; Llaberia Griñó, José M. no Despite the effort inverted the last years in commercial compilers to generate efficient SIMD instructions based code sequences from conventional sequential programs, the small numbers of compilers that can automatically use these instructions achieve in most cases unsatisfactory results. This work shows how exposing register level reuse in source codes helps vectorizing compilers as ICC to generate efficient SIMD code that exploits vectorial register reuse. To this end we use reverse-engineering to know the limitations of ICC compiler. We compare performance of compiler generated code to performance of hand-optimized assembly-written numerical libraries achieving around the 72% performance of MKL for example. http://hdl.handle.net/2117/16301 Title: ICT4Girls: from high school to university: an approach Authors: Gil, Marisa; Jiménez Castells, Marta; Otero Calviño, Beatriz Abstract: Technological advances are improving living conditions. Oddly, there continues to be a decrease in the number of ICT students. The decline in the number of female students is particularly noteworthy. One of the main reasons is that neither the model of professional profile promoted, nor the global approach of the mission is attractive or appealing. In this paper we present our ICT4Girls project with two goals: to collaborate with high schools in mentoring and encouraging students (mainly girls) to integrate their skills with their personal vocation and their future career; and to present computer science as a real tool for providing service to the society, promoting ICT careers and making women visible in the world of technology. The results and conclusions we present are based on three activities carried out in a pilot project. Thu, 19 Jul 2012 10:37:07 GMT http://hdl.handle.net/2117/16301 2012-07-19T10:37:07Z Gil, Marisa; Jiménez Castells, Marta; Otero Calviño, Beatriz no ICT, Profession, Society, Visibility of women, Social impact Technological advances are improving living conditions. Oddly, there continues to be a decrease in the number of ICT students. The decline in the number of female students is particularly noteworthy. One of the main reasons is that neither the model of professional profile promoted, nor the global approach of the mission is attractive or appealing. In this paper we present our ICT4Girls project with two goals: to collaborate with high schools in mentoring and encouraging students (mainly girls) to integrate their skills with their personal vocation and their future career; and to present computer science as a real tool for providing service to the society, promoting ICT careers and making women visible in the world of technology. The results and conclusions we present are based on three activities carried out in a pilot project. http://hdl.handle.net/2117/15944 Title: Introducing risk management into cloud computing Authors: Fitó, Josep Oriol; Guitart Fernández, Jordi Abstract: The Cloud computing paradigm is offering an innovative and promising vision concerning the Information and Communications Technology (ICT). Actually, it gives the possibility of improving IT systems management and is changing the way in which hardware and software are designed and purchased. Notwithstanding, the use of Cloud resources, which usually are external assets to their consumers, implies risk issues that must be taken into account. In this paper, we propose the involvement of risk management procedures into Cloud computing. In this sense, we present a Cloud computing risk management approach aware of Business- Level Objectives (BLOs) of a given Cloud organization. More to the point, we propose an innovatory SEmi-quantitative BLOdriven Cloud Risk Assessment (SEBCRA) as the core subprocess of this Cloud risk management approach. In addition, we present, as a use case, a Cloud Service Provider (CSP) that is able to improve the achievement of a given BLO, i.e. profit maximization, by managing, assessing, and treating Cloud-related risks. As demonstrated in the experimentation, this provider maximizes its profit by transferring the risks of provisioning its private Cloud, either under- or over-provisioning, to third-party Cloud Infrastructure Providers (CIPs). Wed, 30 May 2012 09:44:59 GMT http://hdl.handle.net/2117/15944 2012-05-30T09:44:59Z Fitó, Josep Oriol; Guitart Fernández, Jordi no Semi-quantitative risk management, Business-level objectives, Cloud service providers The Cloud computing paradigm is offering an innovative and promising vision concerning the Information and Communications Technology (ICT). Actually, it gives the possibility of improving IT systems management and is changing the way in which hardware and software are designed and purchased. Notwithstanding, the use of Cloud resources, which usually are external assets to their consumers, implies risk issues that must be taken into account. In this paper, we propose the involvement of risk management procedures into Cloud computing. In this sense, we present a Cloud computing risk management approach aware of Business- Level Objectives (BLOs) of a given Cloud organization. More to the point, we propose an innovatory SEmi-quantitative BLOdriven Cloud Risk Assessment (SEBCRA) as the core subprocess of this Cloud risk management approach. In addition, we present, as a use case, a Cloud Service Provider (CSP) that is able to improve the achievement of a given BLO, i.e. profit maximization, by managing, assessing, and treating Cloud-related risks. As demonstrated in the experimentation, this provider maximizes its profit by transferring the risks of provisioning its private Cloud, either under- or over-provisioning, to third-party Cloud Infrastructure Providers (CIPs). http://hdl.handle.net/2117/15943 Title: Maximising revenue in cloud computing markets by means of economically enhanced SLA management Authors: Macías Lloret, Mario; Guitart Fernández, Jordi Abstract: This paper proposes a bidirectional communication between market brokers and resource managers in Cloud Computing Markets. This communication is implemented by means of an Economically Enhanced Resource Manager (EERM), that supports the negotiation process by deciding which tasks can be allocated or not, and under which economic and technical conditions. The EERM also uses the economic information that collects from market layers to manage the resources accordingly to concrete BLOs. This paper shows several Business Policies and Rules for maximizing the revenue of a Cloud Provider that sells its services and resources in a market. Their validity is demonstrated through several experiments that shown how the application of these rules can have a positive influence in the revenue and minimize the violations of Service-Level Agreements. Wed, 30 May 2012 09:32:24 GMT http://hdl.handle.net/2117/15943 2012-05-30T09:32:24Z Macías Lloret, Mario; Guitart Fernández, Jordi no Markets, Economic Resource Management This paper proposes a bidirectional communication between market brokers and resource managers in Cloud Computing Markets. This communication is implemented by means of an Economically Enhanced Resource Manager (EERM), that supports the negotiation process by deciding which tasks can be allocated or not, and under which economic and technical conditions. The EERM also uses the economic information that collects from market layers to manage the resources accordingly to concrete BLOs. This paper shows several Business Policies and Rules for maximizing the revenue of a Cloud Provider that sells its services and resources in a market. Their validity is demonstrated through several experiments that shown how the application of these rules can have a positive influence in the revenue and minimize the violations of Service-Level Agreements. http://hdl.handle.net/2117/15230 Title: Estudio y evaluación de formatos de almacenamiento para matrices dispersas en arquitecturas multi-core Authors: Pasarin, Marc; Otero Calviño, Beatriz; Herrero Zaragoza, José Ramón Fri, 17 Feb 2012 18:37:51 GMT http://hdl.handle.net/2117/15230 2012-02-17T18:37:51Z Pasarin, Marc; Otero Calviño, Beatriz; Herrero Zaragoza, José Ramón no http://hdl.handle.net/2117/12320 Title: Escolta al teu voltant, xarxes de sensors sense fils Authors: Crusafon, Carlota; Carcelle, Nuria; Punsoda, Georgina; Gil, Marisa Abstract: Escolta al teu voltant! és un projecte que consisteix en desenvolupar una plataforma de monitorització utilitzant una xarxa de sensors sense fils formada per motes. Una mote és un mòdul de baixa potència que incorpora un microprocessador i varis sensors a més d’una antena de radio. Els sensors permeten obtenir informació de l’entorn mentre que el microprocessador ens permet gestionar la informació obtingua i tranformar-la en missatges que puguin ser enviats a altres motes mitjançant la radio. L’objectiu principal del nostre projecte és proporcionar la informació necessària per sensibilitzar a la comunitat de l’estalvi energètic, millorant el hàbits de les persones en quan als serveis de calefacció, refrigeració i il.luminació. Ens centrarem en el Campus Nord de la UPC, per tal d’afavorir una universitat sostenible. El propi projecte ha de ser ell mateix sotenible i per tant escollirem sempre les opcions que suposin menys energia i siguin menys agressives, tant amb les persones com amb el medi ambient. Hi participen estudiants i professors de la FIB amb la col.laboració de l’empresa DEXMA. Description: Technical Report Fri, 08 Apr 2011 11:09:25 GMT http://hdl.handle.net/2117/12320 2011-04-08T11:09:25Z Crusafon, Carlota; Carcelle, Nuria; Punsoda, Georgina; Gil, Marisa no Escolta al teu voltant! és un projecte que consisteix en desenvolupar una plataforma de monitorització utilitzant una xarxa de sensors sense fils formada per motes. Una mote és un mòdul de baixa potència que incorpora un microprocessador i varis sensors a més d’una antena de radio. Els sensors permeten obtenir informació de l’entorn mentre que el microprocessador ens permet gestionar la informació obtingua i tranformar-la en missatges que puguin ser enviats a altres motes mitjançant la radio. L’objectiu principal del nostre projecte és proporcionar la informació necessària per sensibilitzar a la comunitat de l’estalvi energètic, millorant el hàbits de les persones en quan als serveis de calefacció, refrigeració i il.luminació. Ens centrarem en el Campus Nord de la UPC, per tal d’afavorir una universitat sostenible. El propi projecte ha de ser ell mateix sotenible i per tant escollirem sempre les opcions que suposin menys energia i siguin menys agressives, tant amb les persones com amb el medi ambient. Hi participen estudiants i professors de la FIB amb la col.laboració de l’empresa DEXMA. http://hdl.handle.net/2117/11061 Title: An integer linear programming representation for data-center power-aware management Authors: Berral García, Josep Lluís; Gavaldà Mestre, Ricard; Torres Viñals, Jordi Abstract: This work exposes how to represent a grid data-center based scheduling problem, taking the advantages of the virtualization and consolidation techniques, as a linear integer programming problem including all three mentioned factors. Although being integer linear programming (ILP) a computationally hard problem, specifying correctly its constraints and optimization function can contribute to find integer optimal solutions in relative short time. So ILP solutions can help designers and system managers not only to apply them to schedulers but also to create new heuristics and holistic functions that approximate well to the optimal solutions in a quicker way. Mon, 17 Jan 2011 11:21:12 GMT http://hdl.handle.net/2117/11061 2011-01-17T11:21:12Z Berral García, Josep Lluís; Gavaldà Mestre, Ricard; Torres Viñals, Jordi no This work exposes how to represent a grid data-center based scheduling problem, taking the advantages of the virtualization and consolidation techniques, as a linear integer programming problem including all three mentioned factors. Although being integer linear programming (ILP) a computationally hard problem, specifying correctly its constraints and optimization function can contribute to find integer optimal solutions in relative short time. So ILP solutions can help designers and system managers not only to apply them to schedulers but also to create new heuristics and holistic functions that approximate well to the optimal solutions in a quicker way. http://hdl.handle.net/2117/7403 Title: El optimizador de bucles del compilador Open64/ORC Authors: Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. Fri, 28 May 2010 09:08:23 GMT http://hdl.handle.net/2117/7403 2010-05-28T09:08:23Z Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. no http://hdl.handle.net/2117/7397 Title: Introducción al compilador Open64/ORC Authors: Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. Thu, 27 May 2010 12:23:38 GMT http://hdl.handle.net/2117/7397 2010-05-27T12:23:38Z Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. no http://hdl.handle.net/2117/7395 Title: El optimizador de bucles del compilador Open64/ORC (parte 2) Authors: Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. Abstract: Open64 y ORC (Open Research Compiler) son dos iniciativas de código abierto basadas en el compilador SGI Pro64. Open64 está gestionada por miembros de la Universidad de Delaware, y ORC es una extensión del compilador desarrollada por Intel y la Chinese Academy of Science. Para más información consultar las respectivas páginas web [2] y [1]. SGI Pro64 es un conjunto de compiladores optimizadores desarrollados por SGI. Incluye compiladores de C, C++ y Fortran90/95 que siguen los estándares ABI y API de Linux IA-64. Los archivos fuente son de dominio público y se distribuyen bajo los términos de la GNU General Public License. El conjunto de compiladores está disponible para correr sobre plataformas Linux IA-32 e IA-64. Este documento continúa el trabajo iniciado en los technical reports “Introducción al compilador Open64/ORC” [10] y “El optimizador de bucles del compilador Open64/ORC (parte 1)” [11]. El primero describe los componentes del compilador y la representación intermedia que se utiliza como interficie común entre ellos. El segundo documento se centra específicamente en uno de los componentes del compilador: el optimizador de bucles. Thu, 27 May 2010 11:49:29 GMT http://hdl.handle.net/2117/7395 2010-05-27T11:49:29Z Santamaria Barnadas, Eduard; Jiménez Castells, Marta; Fernández Jiménez, Agustín; Llaberia Griñó, José M. no Open64 y ORC (Open Research Compiler) son dos iniciativas de código abierto basadas en el compilador SGI Pro64. Open64 está gestionada por miembros de la Universidad de Delaware, y ORC es una extensión del compilador desarrollada por Intel y la Chinese Academy of Science. Para más información consultar las respectivas páginas web [2] y [1]. SGI Pro64 es un conjunto de compiladores optimizadores desarrollados por SGI. Incluye compiladores de C, C++ y Fortran90/95 que siguen los estándares ABI y API de Linux IA-64. Los archivos fuente son de dominio público y se distribuyen bajo los términos de la GNU General Public License. El conjunto de compiladores está disponible para correr sobre plataformas Linux IA-32 e IA-64. Este documento continúa el trabajo iniciado en los technical reports “Introducción al compilador Open64/ORC” [10] y “El optimizador de bucles del compilador Open64/ORC (parte 1)” [11]. El primero describe los componentes del compilador y la representación intermedia que se utiliza como interficie común entre ellos. El segundo documento se centra específicamente en uno de los componentes del compilador: el optimizador de bucles. http://hdl.handle.net/2117/7138 Title: Deadline constrained prediction of job resource requirements to manage high-level SLAs for SaaS cloud providers Authors: Reig Ventura, Gemma; Alonso López, Javier; Guitart Fernández, Jordi Abstract: For a non IT expert to use services in the Cloud is more natural to negotiate the QoS with the provider in terms of service-level metrics –e.g. job deadlines– instead of resourcelevel metrics –e.g. CPU MHz. However, current infrastructures only support resource-level metrics –e.g. CPU share and memory allocation– and there is not a well-known mechanism to translate from service-level metrics to resource-level metrics. Moreover, the lack of precise information regarding the requirements of the services leads to an inefficient resource allocation –usually, providers allocate whole resources to prevent SLA violations. According to this, we propose a novel mechanism to overcome this translation problem using an online prediction system which includes a fast analytical predictor and an adaptive machine learning based predictor. We also show how a deadline scheduler could use these predictions to help providers to make the most of their resources. Our evaluation shows: i) that fast algorithms are able to make predictions with an 11% and 17% of relative error for the CPU and memory respectively; ii) the potential of using accurate predictions in the scheduling compared to simple yet well-known schedulers. Mon, 10 May 2010 09:43:58 GMT http://hdl.handle.net/2117/7138 2010-05-10T09:43:58Z Reig Ventura, Gemma; Alonso López, Javier; Guitart Fernández, Jordi no For a non IT expert to use services in the Cloud is more natural to negotiate the QoS with the provider in terms of service-level metrics –e.g. job deadlines– instead of resourcelevel metrics –e.g. CPU MHz. However, current infrastructures only support resource-level metrics –e.g. CPU share and memory allocation– and there is not a well-known mechanism to translate from service-level metrics to resource-level metrics. Moreover, the lack of precise information regarding the requirements of the services leads to an inefficient resource allocation –usually, providers allocate whole resources to prevent SLA violations. According to this, we propose a novel mechanism to overcome this translation problem using an online prediction system which includes a fast analytical predictor and an adaptive machine learning based predictor. We also show how a deadline scheduler could use these predictions to help providers to make the most of their resources. Our evaluation shows: i) that fast algorithms are able to make predictions with an 11% and 17% of relative error for the CPU and memory respectively; ii) the potential of using accurate predictions in the scheduling compared to simple yet well-known schedulers. http://hdl.handle.net/2117/6354 Title: Measuring TCP bandwidth on top of a Gigabit and Myrinet network Authors: Costa Prats, Juan José; Bueno Hedo, Javier; Martorell Bofill, Xavier; Cortés Rosselló, Antonio Abstract: In this article we measure the bandwidth achieved by the TCP protocol on top of a Gigabit and Myrinet network. We have created a synthetic benchmark, consisting on a server and a client that both produce and consume data, to measure bandwidth. Four different versions of this benchmark are evaluated: (a) a simple one-threaded version; (b) a version that can consume data if the producer is unable to produce data; (c) a version with the Nagle algorithm disabled; and (d) a version where producers and consumers are threads on their owns. The results shows that on both networks: (1) it is necessary to drain the network to avoid deadlocks; (2) the Nagle algorithm is useful and gets smaller variation in results; and (3) adding an extra thread for reading reduces the maximum bandwidth achievable and it gets better bandwidth for bigger messages and worse for smaller ones when a great number of messages are sent. Fri, 12 Feb 2010 10:51:05 GMT http://hdl.handle.net/2117/6354 2010-02-12T10:51:05Z Costa Prats, Juan José; Bueno Hedo, Javier; Martorell Bofill, Xavier; Cortés Rosselló, Antonio no In this article we measure the bandwidth achieved by the TCP protocol on top of a Gigabit and Myrinet network. We have created a synthetic benchmark, consisting on a server and a client that both produce and consume data, to measure bandwidth. Four different versions of this benchmark are evaluated: (a) a simple one-threaded version; (b) a version that can consume data if the producer is unable to produce data; (c) a version with the Nagle algorithm disabled; and (d) a version where producers and consumers are threads on their owns. The results shows that on both networks: (1) it is necessary to drain the network to avoid deadlocks; (2) the Nagle algorithm is useful and gets smaller variation in results; and (3) adding an extra thread for reading reduces the maximum bandwidth achievable and it gets better bandwidth for bigger messages and worse for smaller ones when a great number of messages are sent.