Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 4
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Solving Weighted Least Squares (WLS) problems on ARM-based architectures

  • Belloch Rodríguez, José Antonio
  • Bank, Balazs
  • Igual Peña, Francisco Daniel
  • Quintana-Ortí, Enrique S.|||0000-0002-5454-165X
  • Vidal Maciá, Antonio Manuel
TheWeighted Least Squares algorithm (WLS) is applied to numerous optimization problems, but requires the use of high computational resources, especially when complex arithmetic is involved. This work aims to accelerate the resolution of a WLS problem by reducing the computational cost (relaying on BLAS/LAPACK routines) and the computational precision from double to single. As a test case, we design an IIR filter for a Graphic Equalizer, where the numerical errors due to single precision are easily visualized. In addition, given the importance of low power architectures for this kind of implementations, we evaluate the performance, scalability, and energy efficiency of each method on two different processors implementing the ARMv7 architecture, widely used in current mobile devices with power constraints. Results show that the method that exhibits a high theoretical computational cost overcomes in efficiency other methods with lower theoretical cost in architectures of this type., This work started in spring 2016 when Jose A. Belloch was a visiting postdoctoral researcher at Budapest University of Technology and Economics thanks to the European Network COST Action IC1305 inside the program Short Term Scientific Mission with the following reference: COST-SPASM-ECOST-STSM-IC1305-020416-072431. Dr. Jose A. Belloch is supported by GVA contract APOSTD/2016/069. The researchers from Universitat Jaume I are supported by the CICYT projects TIN2014-53495-R of MINECO and FEDER. The authors from the Universitat Politecnica de Valencia are supported by MINECO Projects TEC2015-67387-C4-1-R, PROMETEOII/2014/003 and CAPAP-H5 network TIN2014-53522-REDT. The researcher from UCM is supported by the EU (FEDER) and the Spanish MINECO, under Grants TIN 2015-65277-R and TIN2012-32180. The work of Balazs Bank was supported by the UNKP-16-4-III New National Excellence Program of the Ministry of Human Capacities, Hungary.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Exploring Hybrid Parallel Systems for Probabilistic Record Linkage

  • Boratto, Murilo
  • Pinto, Clicia
  • Melo, Pedro
  • Barreto, Marcos
  • Denaxas, Spiros
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] Record linkage is a technique widely used to gather data stored in disparate data sources that presumably pertain to the same real world entity. This integration can be done deterministically or probabilistically, depending on the existence of common key attributes among all data sources involved. The probabilistic approach is very time-consuming due to the amount of records that must be compared, specifically in big data scenarios. In this paper, we propose and evaluate a methodology that simultaneously exploits multicore and multi-GPU architectures in order to perform the probabilistic linkage of large-scale Brazilian governmental databases. We present some algorithmic optimizations that provide high accuracy and improve performance by defining the best algorithm-architecture combination for a problem given its input size. We also discuss performance results obtained with different data samples, showing that a hybrid approach outperforms other configurations, providing an average speedup of 7.9 when linking up to 20.000 million records., This work has been partially supported by CNPq, FAPESB, Bill & Melinda Gates Foundation, The Royal Society (UK), Medical Research Council (UK), NVIDIA Hardware Grant Program, Generalitat Valenciana (Grant PROMETEOII/2014/003), Spanish Government and European Commission through TEC2015-67387-C4-1-R (MINECO/FEDER), and network CAPAP-H. We have also worked in cooperation with the EU-COST Programme Action IC1305, "Network for Sustainable Ultrascale Computing (NESUS)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Artículo científico (JournalArticle)

Automatic Tuning to Performance Modelling of Matrix Polynomials on Multicore and Multi-GPU Systems

  • Boratto, Murilo
  • Gimenez, Domingo
  • Lastovetsky, Alexey
  • Alonso-Jordá, Pedro|||0000-0002-6882-6592
[EN] Automatic tuning methodologies have been used in the design of routines in recent years. The goal of these methodologies is to develop routines which automatically adapt to the conditions of the underlying computational system so that efficient executions are obtained independently of the end- user experience. This paper aims to explore programming routines that can automatically be adapted to the computational system conditions thanks to these automatic tuning methodologies. In particular, we have worked on the evaluation of matrix polynomials on multicore and multi-GPU systems as a target application. This application is very useful for the computation of matrix functions like the sine or cosine but, at the same time, the application is very time consuming since the basic computational kernel, which is the matrix multiplication, is carried out many times. The use of all available resources within a node in an easy and efficient way is crucial for the end user., This work has been partially supported by Generalitat Valenciana under Grant PROM-ETEOII/2014/003, and by the Spanish MINECO, as well as European Commission FEDER funds, under Grant TEC2015-67387-C4-1-R and TIN2015-66972-C5-3-R, and network CAPAP-H. Also, we have work in cooperation with the EU-COST Programme Action IC1305, "Network for Sustainable Ultrascale Computing (NESUS)".
RUC. Repositorio da Universidade da Coruña
Publication Contribución de congreso (ConferenceOutput)

New Tridiagonal Systems Solvers on GPU Architectures

  • Pérez Diéguez, Adrián
  • Amor, Margarita
  • Doallo, Ramón
This version of the article has been accepted for publication, after peer review. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. The Version of Record is available online at: https://doi.org/10.1109/HiPC.2015.17, [Abstract]: Modern GPUs (Graphics Processing Units) offer very high computing power at relatively low cost. Nevertheless, designing efficient algorithms for the GPUs usually requires additional time and effort, even for experienced programmers. On the other hand, tridiagonal systems solvers are an important building block for a wide range of applications. In this paper, we present a new tuning parallel proposal in order to generate new tridiagonal systems solvers. This proposal is based on the combination of a new reduction algorithm (Redundant Reduction-RR) with a tuning proposal to generate efficient parallel prefix algorithms on the GPU. Specifically, we present two new solvers combining RR with two GPU efficient parallel prefix patterns. The performance of the resulting proposals was analyzed using three different CUDA GPUs, obtaining an improvement of up to 20.5x over the CUSPARSE library and 28.9x over CUDPP., This research has been supported by the Galician Government (Xunta de Galicia) under the Consolidation Program of Competitive Reference Groups, cofunded by FEDER funds of the EU (Ref. GRC2013/055); and by the Ministry of Economy and Competitiveness of Spain and FEDER funds of the EU (Project TIN2013-42148-P). This work is also partially supported by EU under the COST Program Action IC1305: Network for Sustainable Ultrascale Computing (NESUS)., Xunta de Galicia; GRC2013/055, Presentado en: 2015 IEEE 22nd International Conference on High Performance Computing (HiPC), 16-19 December 2015