Alberto Proença | Universidade do Minho (original) (raw)

Papers by Alberto Proença

Journal of Computational Science, May 1, 2022

Advances in Parallel & Distributed Processing, and Applications

The generation of large streams of pseudorandom numbers may lead to performance degradation in si... more The generation of large streams of pseudorandom numbers may lead to performance degradation in simulation applications. Both the PRN generator and how it is used impact the efficiency of generating multiple PRNs.

The European Physical Journal Plus

HEP-Frame is a new C++ package designed to efficiently perform analyses of datasets from a very l... more HEP-Frame is a new C++ package designed to efficiently perform analyses of datasets from a very large number of events, like those available at the Large Hadron Collider (LHC) at CERN, Geneva. It mainly targets high-performance servers and mini-clusters, and it was designed for natural science researchers with a user-friendly interface to access structured databases. HEP-Frame automatically evaluates the underlying computing resources and builds an adequate code skeleton when creating a data analysis application. At run-time, HEP-Frame analyses a sequence of datasets exploring the available parallelism in the code and hardware resources: it concurrently reads inputs from a user-defined data structure and processes them, following the user-specific sequence of requirements to select relevant data; it manages the efficient execution of that sequence; and it outputs results in user-defined objects (e.g., ROOT structures), stored together with the used input dataset. This paper shows ho...

This paper uses typed linear algebra (LA) to represent data and perform analytical querying in a ... more This paper uses typed linear algebra (LA) to represent data and perform analytical querying in a single, unified framework. The typed approach offers strong type checking (as in modern programming languages) and a diagrammatic way of expressing queries (paths in LA diagrams). A kernel of LA operators has been implemented so that paths extracted from LA diagrams can be executed. The approach is validated and evaluated taking TPC-H benchmark queries as reference. The performance of the LA-based approach is compared with popular database competitors (PostgreSQL and MySQL).

Proceedings of the Sixth Euromicro Workshop on Parallel and Distributed Processing - PDP '98 -

... parallel class ObjA { public: / / method asynchronously called void aMethod ( ) ; 1; ... / de... more ... parallel class ObjA { public: / / method asynchronously called void aMethod ( ) ; 1; ... / declaration of QbjB void 0bjB::bMethodO { / / declares a pointer to a parallel object parallel ObjA *anObj; / / creates a parallel object of type ...

2015 International Conference on Computational Science and Computational Intelligence (CSCI), 2015

This communication presents an evolutionary software prototype of a user-centered Highly Efficien... more This communication presents an evolutionary software prototype of a user-centered Highly Efficient Pipelined Framework, HEP-Frame, to aid the development of sustainable parallel scientific code with a flexible pipeline structure. HEP-Frame is the result of a tight collaboration between computational scientists and software engineers: it aims to improve scientists coding productivity, ensuring an efficient parallel execution on a wide set of multicore systems, with both HPC and HTC techniques. Current prototype complies with the requirements of an actual scientific code, includes desirable sustainability features and supports at compile time additional plugin interfaces for other scientific fields. The porting and development productivity was assessed and preliminary efficiency results are promising.

Proceedings of the 5th Int. Meeting on High Performance Computing for Computational Science (VECPAR��02), Porto, Portugal, Jun 1, 2002

Abstract. The main role of a scheduler on a distributed environment is to effectively and efficie... more Abstract. The main role of a scheduler on a distributed environment is to effectively and efficiently manage the code and data workload distribution. The scheduling context is presented to identify its main components and their role. To analyse the structure of a generic scheduler three main features are identified and further detailed: the performance model (effectiveness, environment and scheduling overheads metrics), the internal execution model and the decision making mechanism. Deterministic and stochastic ...

Matrix algorithms often deal with large amounts of data at a time, which impairs efficient cache ... more Matrix algorithms often deal with large amounts of data at a time, which impairs efficient cache memory usage. Recent collaborative work between the Numerical Algorithms Group and the University of Minho led to a blocked approach to the matrix square root algorithm with significant efficiency improvements, particularly in a multicore shared memory environment. Distributed memory architectures were left unexplored. In these systems data is distributed across multiple memory spaces, including those associated with specialized accelerator devices, such as GPUs. Systems with these devices are known as heterogeneous platforms. This dissertation focuses on studying the blocked matrix square root algorithm, first in a multicore environment, and then in heterogeneous platforms. Two types of hardware accelerators are explored: Intel Xeon Phi coprocessors and NVIDIA CUDA-enabled GPUs. The initial implementation confirmed the advantages of the blocked method and showed excellent scalability in a multicore environment. The same implementation was also used in the Intel Xeon Phi, but the obtained performance results lagged behind the expected behaviour and the CPU-only alternative. Several optimizations techniques were applied to the common implementation, which managed to reduce the gap between the two environments. The implementation for CUDA-enabled devices followed a different programming model and was not able to benefit from any of the previous solutions. It also required the implementation of BLAS and LAPACK routines, since no existing package fits the requirements of this application. The measured performance also showed that the CPU-only implementation is still the fastest. v Resumo Computação Eficiente da Raíz Quadrada de uma Matriz em Plataformas Heterogéneas Algoritmos de matrizes lidam regularmente com grandes quantidades de dados ao mesmo tempo, o que dificulta uma utilização eficiente da cache. Um trabalho recente de colaboração entre o Numerical Algorithms Group e a Universidade do Minho levou a uma abordagem por blocos para o algoritmo da raíz quadrada de uma matriz com melhorias de eficiência significativas, particularmente num ambiente multicore de memória partilhada. Arquiteturas de memória distribuída permaneceram inexploradas. Nestes sistemas os dados são distribuídos por diversos espaços de memória, incluindo aqueles associados a dispositivos aceleradores especializados, como GPUs. Sistemas com estes dispositivos são conhecidos como plataformas heterogéneas. Esta dissertação foca-se em estudar o algoritmo da raíz quadrada de uma matriz por blocos, primeiro num ambiente multicore e depois usando plataformas heterogéneas. Dois tipos de aceleradores são explorados: co-processadores Intel Xeon Phi e GPUs NVIDIA habilitados para CUDA. A implementação inicial confirmou as vantagens do método por blocos e mostrou uma escalabilidade excelente num ambiente multicore. A mesma implementação foi ainda usada para o Intel Xeon Phi, mas os resultados de performance obtidos ficaram aquém do comportamento esperado e da alternativa usando apenas CPUs. Várias otimizações foram aplicadas a esta implementação comum, conseguindo reduzir a diferença entre os dois ambientes. A implementação para dispositivos CUDA seguiu um modelo de programação diferente e não pôde beneficiar the nenhuma das soluções anteriores. Também exigiu a implementação de rotinas BLAS e LAPACK, já que nenhum dos pacotes existentes se adequa aos requisitos desta implementação. A performance medida também mostrou que a alternativa usando apenas CPUs ainda é a mais rápida. vii

This paper proposes an automatic partition methodology oriented to develop data flow dominated em... more This paper proposes an automatic partition methodology oriented to develop data flow dominated embedded systems. The target architecture is CPU-based with reconfigurable devices on attached board (s), which closely matches the PSM meta-model applied to ...

Computers & Graphics, 2010

The use of serious games and applications has emerged as a dominating force in training, educatio... more The use of serious games and applications has emerged as a dominating force in training, education and simulation due to the focus on creating compelling interactive environments at reduced costs by adopting its capacity of engaging younger hi-tech generations in the highly motivated gaming environments and to the opportunities offered by the commodity technologies commonly associated with the entertainment industries. This field is informed by theories, methods, applications and the state-of-the-art in a number of areas based on ...

VAST, 2007

Reflection Transformation Imaging (RTI) has been successfully used to document cultural heritage ... more Reflection Transformation Imaging (RTI) has been successfully used to document cultural heritage information, namely in the representation of 3D sculptural relief with low depth, using Polynomial Texture Maps (PTM). We developed a set of techniques and a software tool (PTMbuilder) that significantly speeds up the PTM generation through the automatic detection of the lighting directions from a set of previously captured images. PTMbuilder includes an LPtracker module that uses image processing techniques coupled with spatial geometry to find the accurate positions of a glossy sphere placed next to the object and of each highlight on its surface; from these coordinates it computes the direction of the lighting position, as required by a PTM generator. The user can interactively review the computed sphere centre, adjust the light position and check the lights positioning, before proceeding to the generation and viewing of the PTM file through the bundle PTMfitter and PTMviewer modules (HP Labs).

This paper describes a Sensor Observation Service (SOS) client developed to integrate dynamic geo... more This paper describes a Sensor Observation Service (SOS) client developed to integrate dynamic geospatial data from meteorological sensors, on a grid-based risk management decision support system. The present work is part of the CROSS-Fire project, which uses forest fires as the main case study and the FireStation application to simulate fire spread. The meteorological data is accessed through the SOS standard from Open Geospatial Consortium (OGC), using the Observations and Measurements (O&M) standard encoding format. Since the SOS standard was not designed to directly access sensors, we developed an interface application to load the SOS database with observations from a Vantis Weather Station (WS). To integrate the SOS meteorological data into the FireStation, the developed SOS client was embedded on a Web Processing Service (WPS) algorithm. This algorithm was designed to be functional and fully compliant with SOS, SensorML, and O&M standards from OGC. With minor modifications to the developed SOS database interface, the SOS client works with any WS. This client supports spatial and temporal filters, including the integration of dynamic data from satellites into FireStation, as described.

Fire propagation simulation tools are useful at different levels of forest fire management. From ... more Fire propagation simulation tools are useful at different levels of forest fire management. From prescribed fire planning to fuel hazard assessment or to the development of fire suppression strategies on wildfires or even training activities. Nevertheless, real time use of such tools is still very limited among the operational authorities for several reasons: lack of good real time data, lack of training or even lack of confidence on the capabilities of actual systems, among others. Wildfire management is a relevant Civil Protection (CP) activity that involves many different and autonomous actors, from public bodies to research centres and should some how reach the general public as an information and alert system. It requires a fast and reliable risk management support system, with real-time or near real-time availability of critical geo-referenced data and settings-based forecasts for fire spreading. CP applications require a strict integration of human and physical resources that must be shared in a coordinated and effective way and be available for the whole emergency procedure.

The CROSS-Fire project aims to develop a grid-based framework as a risk management decision suppo... more The CROSS-Fire project aims to develop a grid-based framework as a risk management decision support system for the civil protection authorities, using forest fires as the main case study and FireStation as the standalone CAD application that simulates the fire spread over complex topography. The CROSS-Fire initial tasks focussed on the development of a parallel version of the fire simulator engine, P-FireStation, and its porting into the EGEE grid environment, G-FireStation. The main contribution of this communication lies on the definition of an OGC-WS framework to enable a basic set of standard geospatial services that will allow Firestation to be interoperable with standard-based Spatial Data Infrastructures. The P-FireStation version explores the inherent parallel environment offered by clusters at each site of the EGEE grid, to support larger data sets and to improve the accuracy of the predictions. This parallel version relies on the MPI protocol and supports larger data sets taking advantage of the MPI parallel I/O facilities. The G-FireStation version integrates EGEE grid facilities, namely the gLite data management services and tools to access data, the AMGA gLite grid metadata catalogue to manage the simulation I/O data, and the WatchDog tool to monitor and provide data for the interactive control of simulations. CROSS-Fire decision support system will be implemented as a web portal that presents at its core a layer of OGC protocols to access, add and process data: WCS for interoperable access to geospatial coverages, WFS for retrieving geographic features and WPS for geospatial processing services and control of the execution of simulations. The portal is also a connection point for desktop clients with GIS and CAD capabilities. The approach will make G-FireStation more interoperable, allowing to access different data providers and publishing output data for further processing, following the guidelines of the EC CYCLOPS project.

The CROSS-Fire is a Portuguese NGI funded project focusing on the development of a grid-based ris... more The CROSS-Fire is a Portuguese NGI funded project focusing on the development of a grid-based risk management decision support system for the Civil Protection (CP), using forest fires as the main case study. The project defines a general approach for the development of a CP application by defining an architecture that integrates three main layers: the CROSS-Fire Platform and two external infrastructures: a Spatial Data Infrastructure (SDI) and the GRID. The CROSS-Fire Platform is defined and implemented as a set of WPS algorithms dealing with most of the functionalities of its three components: Business Logic, Grid Services and Geospatial services. The present work stresses the relevance of standards adoption: OGC-WS WCS/WFS/WMS/WPS, to exploit/enable geospatial services for data access processing, and OGC-SWE SOS to address other CP data sources, such as meteorological station networks (MSN) or satellites. The adoption of a Web Services (WS) approach allows integrating easily with existing systems typically based on WS technologies. We also present CFS, a grid user interface SDI based client, compliant with OGC and EU INSPIRE directives which allows decision makers to access the spatial data infrastructure, to launch simulations on the grid and visualize the fire propagation simulations.

Journal of Computational Science, May 1, 2022

Advances in Parallel & Distributed Processing, and Applications

The European Physical Journal Plus

Proceedings of the Sixth Euromicro Workshop on Parallel and Distributed Processing - PDP '98 -

2015 International Conference on Computational Science and Computational Intelligence (CSCI), 2015

Proceedings of the 5th Int. Meeting on High Performance Computing for Computational Science (VECPAR��02), Porto, Portugal, Jun 1, 2002

Computers & Graphics, 2010

VAST, 2007

—Lattice-based cryptography methods are well known primitives from the asymmetric cryptographic p... more —Lattice-based cryptography methods are well known primitives from the asymmetric cryptographic primitives set. Although widely studied for decades they have gained new notoriety in the last years because it is believe in the field of cryptography that these methods can withstand attacks by upcoming quantum computers. This paper describes how we optimized the Enumeration algorithm's serial code and then we suggest a new way to parallelize it using PThreads. The results shown in the results section are very promising and show that we can obtain significant performance improvements over the said implementation.

The security of most digital systems is under serious threats due to major technology breakthroug... more The security of most digital systems is under serious threats due to major technology breakthroughs we are experienced in nowadays. Lattice-based cryptosystems are one of the most promising post-quantum types of cryptography, since it is believed to be secure against quantum computer attacks. Their security is based on the hardness of the Shortest Vector Problem and Closest Vector Problem.
Lattice basis reduction algorithms are used in several fields, such as lattice-based cryptography and signal processing. They aim to make the problem easier to solve by obtaining shorter and more orthogonal basis. Some case studies work with numbers with hundreds of digits to ensure harder problems, which require Multiple Precision (MP) arithmetic. This dissertation presents a novel integer representation for MP arithmetic and the algorithms for the associated operations, MpIM. It also compares these implementations with other libraries, such as GNU Multiple Precision Arithmetic Library, where our experimental results display a similar performance and for some operations better performances.
This dissertation also describes a novel lattice basis reduction module, LattBRed, which included a novel efficient implementation of the Qiao’s Jacobi method, a Lenstra-Lenstra-Lovász (LLL) algorithm and associated parallel implementations, a parallel variant of the Block Korkine-Zolotarev (BKZ) algorithm and its implementation and MP versions of the the Qiao’s Jacobi method, the LLL and BKZ algorithms.
Experimental performances measurements with the set of implemented modifications of the Qiao’s Jacobi method show some performance improvements and some degradations but speedups greater than 100 in Ajtai-type bases.