Algorithm-oriented design of efficient many-core architectures applied to dense matrix multiplication

Recent integrated circuit technologies have opened the possibility to design parallel architectures with hundreds of cores on a single chip. The design space of these parallel architectures is huge with many architectural options. Exploring the design space gets even more difficult if, beyond performance and area, we also consider extra metrics like performance and area efficiency, where the designer tries to design the architecture with the best performance per chip area and the best sustainable performance. In this paper we present an algorithm-oriented approach to design a many-core architecture. Instead of doing the design space exploration of the many core architecture based on the experimental execution results of a particular benchmark of algorithms, our approach is to make a formal analysis of the algorithms considering the main architectural aspects and to determine how each particular architectural aspect is related to the performance of the architecture when running an algorithm or set of algorithms. The architectural aspects considered include the number of cores, the local memory available in each core, the communication bandwidth between the many-core architecture and the external memory and the memory hierarchy. To exemplify the approach we did a theoretical analysis of a dense matrix multiplication algorithm and determined an equation that relates the number of execution cycles with the architectural parameters. Based on this equation a many-core architecture has been designed. The results obtained indicate that a 100 mm(2) integrated circuit design of the proposed architecture, using a 65 nm technology, is able to achieve 464 GFLOPs (double precision floating-point) for a memory bandwidth of 16 GB/s. This corresponds to a performance efficiency of 71 %. Considering a 45 nm technology, a 100 mm(2) chip attains 833 GFLOPs which corresponds to 84 % of peak performance These figures are better than those obtained by previous many-core architectures, except for the area efficiency which is limited by the lower memory bandwidth considered. The results achieved are also better than those of previous state-of-the-art many-cores architectures designed specifically to achieve high performance for matrix multiplication.

Matrix converter as unified power flow controller: design and implementation of decoupled direct power controllers

This paper presents the design and implementation of direct power controllers for three-phase matrix converters (MC) operating as Unified Power Flow Controllers (UPFC). Theoretical principles of the decoupled linear power controllers of the MC-UPFC to minimize the cross-coupling between active and reactive power control are established. From the matrix converter based UPFC model with a modified Venturini high frequency PWM modulator, decoupled controllers for the transmission line active (P) and reactive (Q) power direct control are synthesized. Simulation results, obtained from Matlab/Simulink, are presented in order to confirm the proposed approach. Results obtained show decoupled power control, zero error tracking, and fast responses with no overshoot and no steady-state error.

On derivatives and norms of generalized matrix functions and respective symmetric powers

In recent papers, the authors obtained formulas for directional derivatives of all orders, of the immanant and of the m-th xi-symmetric tensor power of an operator and a matrix, when xi is a character of the full symmetric group. The operator norm of these derivatives was also calculated. In this paper, similar results are established for generalized matrix functions and for every symmetric tensor power.

Sparse matrix multiplication on a reconfigurable many-core architecture

Sparse matrix-vector multiplication (SMVM) is a fundamental operation in many scientific and engineering applications. In many cases sparse matrices have thousands of rows and columns where most of the entries are zero, while non-zero data is spread over the matrix. This sparsity of data locality reduces the effectiveness of data cache in general-purpose processors quite reducing their performance efficiency when compared to what is achieved with dense matrix multiplication. In this paper, we propose a parallel processing solution for SMVM in a many-core architecture. The architecture is tested with known benchmarks using a ZYNQ-7020 FPGA. The architecture is scalable in the number of core elements and limited only by the available memory bandwidth. It achieves performance efficiencies up to almost 70% and better performances than previous FPGA designs.

Transverse Vibrations in Beams Supported by a Piece-Wise Homogeneous Visco- Elastic Foundation

Transversal vibrations induced by a load moving uniformly along an infinite beam resting on a piece-wise homogeneous visco-elastic foundation are studied. Special attention is paid to the additional vibrations, conventionally referred to as transition radiations, which arise as the point load traverses the place of foundation discontinuity. The governing equations of the problem are solved by the normalmode analysis. The solution is expressed in a form of infinite sum of orthogonal natural modes multiplied by the generalized coordinate of displacement. The natural frequencies are obtained numerically exploiting the concept of the global dynamic stiffness matrix. This ensures that the frequencies obtained are exact. The methodology has restrictions neither on velocity nor on damping. The approach looks simple, though, the numerical expression of the results is not straightforward. A general procedure for numerical implementation is presented and verified. To illustrate the utility of the methodology parametric optimization is presented and influence of the load mass is studied. The results obtained have direct application in analysis of railway track vibrations induced by high-speed trains when passing regions with significantly different foundation stiffness.

The effect of water inorganic matrix in ibuprofen adsorption onto activated carbon for water and wastewater treatment

Dissertação para obtenção do Grau de Mestre em Engenharia Química e Bioquímica

Spatial neighborhood matrix computation: Inverse distance weighted versus binary contiguity

Paper presented at Geo-Spatial Crossroad GI_Forum, Salzburg, Austria.

An Exact Schedulability Test for Global FP Using State Space Pruning

Presented at 23rd International Conference on Real-Time Networks and Systems (RTNS 2015). 4 to 6, Nov, 2015, Main Track. Lille, France.

Matrix fractional systems

This paper addresses the matrix representation of dynamical systems in the perspective of fractional calculus. Fractional elements and fractional systems are interpreted under the light of the classical Cole–Cole, Davidson–Cole, and Havriliak–Negami heuristic models. Numerical simulations for an electrical circuit enlighten the results for matrix based models and high fractional orders. The conclusions clarify the distinction between fractional elements and fractional systems.

Development of mixed matrix membranes with metal - organic framework and ionic liquids for biogas upgrading

Dissertation presented to Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa for obtaining the master degree in Membrane Engineering

The additive effects in matrix dispersed liquid crystals

Dissertation presented at Faculdade de Ciências e Tecnologia from Universidade Nova de Lisboa to obtain the degree of Master in Chemical and Biochemical Engineering

Effect of polymeric matrix on morphology and photoelectric properties of polymer-quantum dots nanocomposites

Recently, CdTe semiconductor quantum dots (QDs) have attracted great interest due to their unique properties [1]. Their dispersion into polymeric matrices would be very for several optoelectronics applications. Despite its importance, there has been relatively little work done on charge transport in the QD polymeric films [2], which is mainly affected by their structural and morphological properties. In the present work, polymer-quantum dot nanocomposites films based on optically transparent polymers in the visible spectral range and CdTe QDs with controlled particle size and emission wavelength, were prepared via solvent casting. Photoluminescent (PL) measurements indicate different emission intensity of the nanocomposites. A blue shift of the emission peak compared to that of QDs in solution occurred, which is attributed to the QDs environment changes. The morphological and structural properties of the CdTe nanocomposites were evaluated. Since better QDs dispersion was achieved, PMMA seemed to be the most promising matrix. Electrical properties measurements indicate an ohmic behavior.

Thin films composed of gold nanoparticles dispersed in a dielectric matrix: the influence of the host matrix on the optical and mechanical responses

Gold nanoparticles were dispersed in two different dielectric matrices, TiO2 and Al2O3, using magnetron sputtering and a post-deposition annealing treatment. The main goal of the present work was to study how the two different host dielectric matrices, and the resulting microstructure evolution (including both the nanoparticles and the host matrix itself) promoted by thermal annealing, influenced the physical properties of the films. In particular, the structure and morphology of the nanocomposites were correlated with the optical response of the thin films, namely their localized surface plasmon resonance (LSPR) characteristics. Furthermore, and in order to scan the future application of the two thin film system in different types of sensors (namely biological ones), their functional behaviour (hardness and Young's modulus change) was also evaluated. Despite the similar Au concentrations in both matrices (~ 11 at.%), very different microstructural features were observed, which were found to depend strongly on the annealing temperature. The main structural differences included: (i) the early crystallization of the TiO2 host matrix, while the Al2O3 one remained amorphous up to 800 °C; (ii) different grain size evolution behaviours with the annealing temperature, namely an almost linear increase for the Au:TiO2 system (from 3 to 11 nm), and the approximately constant values observed in the Au:Al2O3 system (4–5 nm). The results from the nanoparticle size distributions were also found to be quite sensitive to the surrounding matrix, suggesting different mechanisms for the nanoparticle growth (particle migration and coalescence dominating in TiO2 and Ostwald ripening in Al2O3). These different clustering behaviours induced different transmittance-LSPR responses and a good mechanical stability, which opens the possibility for future use of these nanocomposite thin film systems in some envisaged applications (e.g. LSPR-biosensors).

Development of hyaluronic acid, dextrin and extracellular matrix hydrogels for cell expansion

Dissertação de mestrado integrado em Engenharia Biomédica (área de especialização em Engenharia Clínica)