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.

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.

Monte Carlo methods for matrix computations on the grid

Many scientific and engineering applications involve inverting large matrices or solving systems of linear algebraic equations. Solving these problems with proven algorithms for direct methods can take very long to compute, as they depend on the size of the matrix. The computational complexity of the stochastic Monte Carlo methods depends only on the number of chains and the length of those chains. The computing power needed by inherently parallel Monte Carlo methods can be satisfied very efficiently by distributed computing technologies such as Grid computing. In this paper we show how a load balanced Monte Carlo method for computing the inverse of a dense matrix can be constructed, show how the method can be implemented on the Grid, and demonstrate how efficiently the method scales on multiple processors. (C) 2007 Elsevier B.V. All rights reserved.

Alternatives for parallel Krylov subspace basis computation

Numerical methods related to Krylov subspaces are widely used in large sparse numerical linear algebra. Vectors in these subspaces are manipulated via their representation onto orthonormal bases. Nowadays, on serial computers, the method of Arnoldi is considered as a reliable technique for constructing such bases. However, although easily parallelizable, this technique is not as scalable as expected for communications. In this work we examine alternative methods aimed at overcoming this drawback. Since they retrieve upon completion the same information as Arnoldi's algorithm does, they enable us to design a wide family of stable and scalable Krylov approximation methods for various parallel environments. We present timing results obtained from their implementation on two distributed-memory multiprocessor supercomputers: the Intel Paragon and the IBM Scalable POWERparallel SP2. (C) 1997 by John Wiley & Sons, Ltd.

Subspace wavepacket evolution with Newton polynomials

Time-dependent wavepacket evolution techniques demand the action of the propagator, exp(-iHt/(h)over-bar), on a suitable initial wavepacket. When a complex absorbing potential is added to the Hamiltonian for combating unwanted reflection effects, polynomial expansions of the propagator are selected on their ability to cope with non-Hermiticity. An efficient subspace implementation of the Newton polynomial expansion scheme that requires fewer dense matrix-vector multiplications than its grid-based counterpart has been devised. Performance improvements are illustrated with some benchmark one and two-dimensional examples. (C) 2001 Elsevier Science B.V. All rights reserved.

Análisis de prestaciones de procesadores multi-core y multi-thread

Los procesadores multi-core y el multi-threading por hardware permiten aumentar el rendimiento de las aplicaciones. Por un lado, los procesadores multi-core combinan 2 o más procesadores en un mismo chip. Por otro lado, el multi-threading por hardware es una técnica que incrementa la utilización de los recursos del procesador. Este trabajo presenta un análisis de rendimiento de los resultados obtenidos en dos aplicaciones, multiplicación de matrices densas y transformada rápida de Fourier. Ambas aplicaciones se han ejecutado en arquitecturas multi-core que explotan el paralelismo a nivel de thread pero con un modelo de multi-threading diferente. Los resultados obtenidos muestran la importancia de entender y saber analizar el efecto del multi-core y multi-threading en el rendimiento.

Ultrastructural lesions in the myocardium and kidneys of rabbits in experimental acute Amorimia exotropica poisoning

Abstract: Amorimia exotropica is an important plant associated with sudden death in cattle in Southern Brazil. In order to understand the mechanisms by which A. exotropica causes acute lesions in the heart and kidney of intoxicated animals, an experiment was conducted to determine the histopathology and ultrastructure of myocardial and renal lesions of intoxicated rabbits. After receiving 18g/kg of dried plant, six rabbits died suddenly. At necropsy, the liver was swollen and no other macroscopic lesions were observed. Histologically, centrolobular and midzonal hepatocytes were vacuolated. These vacuoles were strong PAS stained positive, suggesting that they corresponded to glycogen accumulations. In some regions of the ventricular septum and ventricles were found vacuoles of different sizes and the kidneys of two rabbits showed vacuolar degeneration on distal convoluted tubules. Ultrastructurally, the myocardium had cardiomyocytes swelling with separation of myofibrils bundles and rupture and disorganization of the sarcomeres. The mitochondria displayed swelling, disorganization, disruption of the mitochondrial cristae, and electron-dense matrix. Some mitochondria exhibited eccentric projections of their membranes with disruption of both outer and inner membranes. The sarcoplasmic reticulum had no alterations, whereas the T-tubule system was occasionally dilated and ruptured. The kidneys had mitochondrial swelling with disorganization and disruption of the mitochondrial cristae. The vacuoles result from the swelling of the endoplasmatic reticulum and usually were located between two basolateral infoldings and mitochondria, occurring preferentially around the nucleus. The myocytes and T system damages induced by A. exotropica result in acute heart failure and death. Furthermore, this mechanism of cardiotoxicity may be common to all plant containing monofluoroacetate.

Método de inversão matricial rápido GSGR para vectoring em sistemas xDSL, com implementação em chip DSP

Este trabalho apresenta um método rápido de inversão de matrizes densas, e uma possível aplicação com métodos de Vectoring, em pré-codificação e cancelamento de crosstalk de sistemas xDSL. A família de tecnologias xDSL utiliza os pares trançados de fios de cobre telefônicos como meio físico para transmitir dados digitais. O crosstalk é a principal causa de degradação de sinais na mais nova geração de sistemas xDSL, o G.fast, e para combatê-lo são utilizadas técnicas de pré-codificação e cancelamento, chamadas de Vectoring. O método proposto, chamado de GSGR, consiste em uma abordagem diferente para o método clássico de Squared Givens Rotations (SGR), adequado a implementações em plataformas embarcadas de processamento digital de sinais. Foram realizados testes comparativos do método GSGR com métodos diretos clássicos de inversão, utilizando uma plataforma digital multicore baseada no chip TI DSP TMS320C6670 e a plataforma de software Matlab. Os resultados dos testes de inversão de matrizes usando dados reais e dados simulados mostraram que o GSGR foi superior em velocidade de execução sem apresentar perdas significativas de acurácia para a aplicação em sistemas xDSL.

Aspectos ultraestruturais do espermatozóide de Natica marochiensish (Gmelin) (Mollusca, Gastropoda) do litoral norte do Brasil

Spermatozoa of Natica marochienssish (Gmelin, 1791) is described by light and electron microscopy. The spermatozoon is of the primitive type with head contains a conical acrosomal complex with an acrosomal vesicle of dense matrix having a basis occupied by the subacrosomal space. The middle piece shows the centriolar complex surrounded by mitochondria and the tail contains the axoneme with a 9+2.

Side Effects of Neem Oil on the Midgut Endocrine Cells of the Green Lacewing Ceraeochrysa claveri (Navas) (Neuroptera: Chrysopidae)

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Organic protomolecule assembly in igneous minerals

C—H stretching bands, νCH, in the infrared spectrum of single crystals of nominally high purity, of laboratory-grown MgO, and of natural upper mantle olivine, provide an “organic” signature that closely resembles the symmetrical and asymmetrical C—H stretching modes of aliphatic —CH2 units. The νCH bands indicate that H2O and CO2, dissolved in the matrix of these minerals, converted to form H2 and chemically reduced C, which in turn formed C—H entities, probably through segregation into defects such as dislocations. Heating causes the C—H bonds to pyrolyze and the νCH bands to disappear, but annealing at 70°C causes them to reappear within a few days or weeks. Modeling dislocations in MgO suggests that the segregation of C can lead to Cx chains, x = 4, with the terminal C atoms anchored to the MgO matrix by bonding to two O−. Allowing H2 to react with such Cx chains leads to [O2C(CH2)2CO2] or similar precipitates. It is suggested that such Cx—Hy—Oz entities represent protomolecules from which derive the short-chain carboxylic and dicarboxylic and the medium-chain fatty acids that have been solvent-extracted from crushed MgO and olivine single crystals, respectively. Thus, it appears that the hard, dense matrix of igneous minerals represents a medium in which protomolecular units can be assembled. During weathering of rocks, the protomolecular units turn into complex organic molecules. These processes may have provided stereochemically constrained organics to the early Earth that were crucial to the emergence of life.

Mechanism of adhesion and detachment at the anterior end of Neoheterocotyle rhinobatidis and Troglocephalus rhinobatidis (Monogenea : Monopisthocotylea : Monocotylidae)

The anterior adhesion and detachment mechanisms observed for Neoheterocotyle rhinobatidis and Troglocephalus rhinobatidis (Monogenea: Monocotylidae) appear similar to those observed for the two other monopisthocotylean monogenean species with anterior apertures for which published data are available. This supports the theory that monogeneans with apertures may utilise a common mechanism. Adult anterior apertures can open and close and duct endings can evert during the adhesion phase and retract during detachment and searching behaviour. The adhesive is comprised of two secretory types, rod-shaped and spheroidal bodies, found within anterior apertures. These exit together and undergo mixing to produce the adhesive matrix in which elongate membranes from rod-shaped bodies are seen intermixed with a granular electron-dense matrix. The morphology of the adhesive matrix differs from that found for some other monogenean taxa. Anterior detachment by these monocotylids appears to involve a depletion of rod-shaped bodies in ducts and mechanical withdrawal of the anterior end.

A Lanczos-powered implementation of the Faber polynomial quantum time propagator for reaction probabilities

Complementing our recent work on subspace wavepacket propagation [Chem. Phys. Lett. 336 (2001) 149], we introduce a Lanczos-based implementation of the Faber polynomial quantum long-time propagator. The original version [J. Chem. Phys. 101 (1994) 10493] implicitly handles non-Hermitian Hamiltonians, that is, those perturbed by imaginary absorbing potentials to handle unwanted reflection effects. However, like many wavepacket propagation schemes, it encounters a bottleneck associated with dense matrix-vector multiplications. Our implementation seeks to reduce the quantity of such costly operations without sacrificing numerical accuracy. For some benchmark scattering problems, our approach compares favourably with the original. (C) 2004 Elsevier B.V. All rights reserved.

Mitochondrial dysmorphology in the neuroepithelium of rat embryos following a single dose of maternal hyperthermia during gestation

Hyperthermia is teratogenic to human and animal embryos and induces mainly anomalies of the nervous system. However, the teratogenic mechanism is poorly understood. Mammalian embryos are known to switch from anaerobic to aerobic metabolism around the time of neural tube closure. This critical event might be sensitive to hyperthermia. The objective of the present study was to evaluate the ultrastructural changes of the mitochondria of the neuroepithelium (NE) of rat embryos following maternal exposure to hyperthermia. Pregnant rats were heat stressed for an hour on gestation day (GD) 9 and embryos were examined by electron microscopy on GD 10. NE presented extensive apoptosis. Intercellular junctions were weakened and copious cellular debris projected into the ventricle. The mitochondria were of diverse size and shape. Most of them were swollen and had short cristae and electron dense matrix. Hydropic changes were also observed in numerous mitochondria. Lipid-laden mitochondria were found in the apical portions of neuroblasts. The mesenchyme (ME) of heat-treated embryos showed paucity of cells and only as frequent apoptosis as the controls. Their mitochondria also showed changes similar to those of the NE. Additionally extensive lipid accumulation was observed in and in the vicinity of mitochondria, often surrounded by short strands of endoplasmic reticulum. Whereas mitochondrial pathology was associated with profound apoptosis in the NE, growth restriction and lipid accumulation accompanied mitochondrial changes in the ME. The results of this study indicate that the embryonic response to maternal heat shock is tissue-specific and morphologically distinct in this species.

Microenvironment and prognostic factors in bone tumors: IDH mutations in chondrosarcoma

Microenvironment in bone tumors is a dynamic entity composed of cells from different origins (immune cells, stromal cells, mesenchymal stem cells, endothelial cells, pericytes) and vascular structures surrounded by a matrix of different nature (bone, cartilage, myxoid). Interactions between cancer cells and tumor microenvironment (TME) are complex and can change as tumor progress, but are also crucial in determining response to cancer therapies. Chondrosarcoma is the second most frequent bone cancer in adult age, but its treatment still represents a challenge, for the intrinsic resistance to conventional chemotherapy and radiation therapy. This resistance is mainly due to pathological features, as dense matrix, scarce mitoses and poor vascularization, sustained by biological mechanisms only partially delucidated. Somatic mutation in the Krebs cycle enzyme isocytrate dehydrogenase (IDH) have been described in gliomas, acute myeloid leukemia, cholangiocarcinoma, melanoma, colorectal, prostate cancer, thyroid carcinoma and other cancers. In mesenchymal tumors IDH mutations are present in about 50% of central chondrosarcoma. IDH mutations are an early event in chondrosarcoma-genesis, and contribute to the acquisition of malignancy through the block of cellular differentiation, hypoxia induction through HIF stabilization, DNA methylation and alteration of cellular red-ox balance. While in gliomas IDH mutations confers a good prognosis, in chondrosarcoma IDH prognostic role is controversial in different reported series. First aim of this project is to define the prevalence and the prognostic role of IDH mutation in high grade central conventional chondrosarcoma patients treated at Istituto Ortopedico Rizzoli. Second aim is the critical revision of scientific literature to understand better how a genomic event in cancer cell can trigger alteration in the TME, through immune infiltrate reshaping, angiogenesis induction, metabolic and methylation rewiring. Third aim is to screen other sarcoma histotypes for the presence of IDH mutation.