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.

Determinantes de la localización del empleo urbano en Bogotá, Colombia

The spatial dimension appears as a fundamental to board the urban employment generation topics. The space plays a key role on the location decisions made by the agents due to not all places offer the same levels of utility and profitability. The existence of a number of advantages in certain places such as the agglomeration, economies of scale, variety, and accessibility among others, may contribute to explain why and where urban employment is generated. From a theoretical model of preferences for variety, data analysis space exploration (ESDA) and spatial econometrics, it is shown how such spatial advantages can affect the employment generation using Bogot´a as a casestudy.

The mobility system of the multi-tasking rover (MTR)

Current and planned robotic rovers for space exploration are focused on science and correspondingly carry a science payload. Future missions will need robotic rovers that can demonstrate a wider range of functionality. This paper proposes an approach to offering this greater functionality by employing science and/or tool packs aboard a highly mobile robotic chassis. The packs are interchangeable and each contains different instruments or tools. The appropriate selection of science and/or tool packs enables the robot to perform a great variety of tasks either alone or in cooperation with other robots. The multi-tasking rover (MTR), thus conceived, provides a novel method for high return on investment. This paper describes the mobility system of the MTR and reports on initial experimental evaluation of the robotic chassis.

The domain layer for mixed-initiative interaction in generative design

The design space exploration formalism has developed data structures and algorithms of sufficient complexity and scope to support conceptual layout, massing, and enclosure configurations. However, design remains a human enterprise. To support the user in designing with the formalism, we have developed an interaction model that addresses the interleaving of user actions with the formal operations of design space exploration. The central feature of our interaction model is the modeling of control based on mixed-initiative. Initiative is sometimes taken by the designer and sometimes by the formalism in working on a shared design task. The model comprises three layers, domain, task, and dialogue. In this paper we describe the formulation of the domain layer of our mixed-initiative interaction model for design space exploration. We present the view of the domain as understood in the formalism in terms of the three abstract concepts of state, move, and structure. In order to support mixed initiative, it is necessary to develop a shared view of the domain. The domain layer addresses this problem by mapping the designer's view onto the symbol substrate. First, we present the designer's view of the domain in terms of problems, solutions, choices, and history. Second, we show how this view is interleaved with the symbol-substrate through four domain layer constructs, problem state, solution state, choice, and exploration history. The domain layer presents a suitable foundation for integrating the role of the designer with a description formalism. It enables the designer to maintain exploration freedom in terms of formulating and reformulating problems, generating solutions, making choices, and navigating the history of exploration.

Experiments with stochastic processes : facade subdivision based on wind motion

Constraint based tools for architectural design exploration need to satisfy aesthetic and functional criteria as well as combine discrete and continuous modes of exploration. In this paper, we examine the possibilities for stochastic processes in design space exploration.

Specifically, we address the application of a stochastic wind motion model to the subdivision of an external building envelope into smaller discrete components. Instead of deterministic subdivision constraints, we introduce explicit uncertainty into the system of subdivision. To address these aims, we develop a model of stochastic wind motion; create a subdivision scheme that is governed by the wind model and explore a design space of a facade subdivision problem. A discrete version of the facade, composed of light strips and panels, based on the bamboo elements deformed by continuous wind motion, is developed. The results of the experiments are presented in the paper.

Alien

Alien is an instrumental piece using warped synth sounds and evoking science fiction types of imagery. The idea behind this piece of audio was to create an unnerving presence with the use of spoken vocals passed through a filter to deepen and slow the voice.

Alien wailing cinematic score

A short composition featuring unusual soundscapes to create an eerie cinematic piece suitable for science fiction styled projects.

Intelligent machines : an introduction

In this chapter, an introduction to intelligent machine is presented. An explanation on intelligent behavior, and the difference between intelligent and repetitive natural or programmed behavior is provided. Some learning techniques in the field of Artificial Intelligence in constructing intelligent machines are then discussed. In addition, applications of intelligent machines to a number of areas including aerial navigation, ocean and space exploration, and humanoid robots are presented.

Reuse-based test planning for core-based systems-on-chip

Electronic applications are currently developed under the reuse-based paradigm. This design methodology presents several advantages for the reduction of the design complexity, but brings new challenges for the test of the final circuit. The access to embedded cores, the integration of several test methods, and the optimization of the several cost factors are just a few of the several problems that need to be tackled during test planning. Within this context, this thesis proposes two test planning approaches that aim at reducing the test costs of a core-based system by means of hardware reuse and integration of the test planning into the design flow. The first approach considers systems whose cores are connected directly or through a functional bus. The test planning method consists of a comprehensive model that includes the definition of a multi-mode access mechanism inside the chip and a search algorithm for the exploration of the design space. The access mechanism model considers the reuse of functional connections as well as partial test buses, cores transparency, and other bypass modes. The test schedule is defined in conjunction with the access mechanism so that good trade-offs among the costs of pins, area, and test time can be sought. Furthermore, system power constraints are also considered. This expansion of concerns makes it possible an efficient, yet fine-grained search, in the huge design space of a reuse-based environment. Experimental results clearly show the variety of trade-offs that can be explored using the proposed model, and its effectiveness on optimizing the system test plan. Networks-on-chip are likely to become the main communication platform of systemson- chip. Thus, the second approach presented in this work proposes the reuse of the on-chip network for the test of the cores embedded into the systems that use this communication platform. A power-aware test scheduling algorithm aiming at exploiting the network characteristics to minimize the system test time is presented. The reuse strategy is evaluated considering a number of system configurations, such as different positions of the cores in the network, power consumption constraints and number of interfaces with the tester. Experimental results show that the parallelization capability of the network can be exploited to reduce the system test time, whereas area and pin overhead are strongly minimized. In this manuscript, the main problems of the test of core-based systems are firstly identified and the current solutions are discussed. The problems being tackled by this thesis are then listed and the test planning approaches are detailed. Both test planning techniques are validated for the recently released ITC’02 SoC Test Benchmarks, and further compared to other test planning methods of the literature. This comparison confirms the efficiency of the proposed methods.

A Modular and digitally programmable interface based on band-pass sigma-delta modulator for mixed-signal systems-on-chip

The focus of this thesis is to discuss the development and modeling of an interface architecture to be employed for interfacing analog signals in mixed-signal SOC. We claim that the approach that is going to be presented is able to achieve wide frequency range, and covers a large range of applications with constant performance, allied to digital configuration compatibility. Our primary assumptions are to use a fixed analog block and to promote application configurability in the digital domain, which leads to a mixed-signal interface. The use of a fixed analog block avoids the performance loss common to configurable analog blocks. The usage of configurability on the digital domain makes possible the use of all existing tools for high level design, simulation and synthesis to implement the target application, with very good performance prediction. The proposed approach utilizes the concept of frequency translation (mixing) of the input signal followed by its conversion to the ΣΔ domain, which makes possible the use of a fairly constant analog block, and also, a uniform treatment of input signal from DC to high frequencies. The programmability is performed in the ΣΔ digital domain where performance can be closely achieved according to application specification. The interface performance theoretical and simulation model are developed for design space exploration and for physical design support. Two prototypes are built and characterized to validate the proposed model and to implement some application examples. The usage of this interface as a multi-band parametric ADC and as a two channels analog multiplier and adder are shown. The multi-channel analog interface architecture is also presented. The characterization measurements support the main advantages of the approach proposed.

Estudo multitemporal da dinâmica espacial do estuário Potengi/RN, utilizando como base os dados de Sensoriamento Remoto.

This work was developed in the Potengi river estuary, northern coastal city of Natal, located in the State of Rio Grande do Norte. The objective was to study the dynamics of Multitemporal Space in that estuary, however, analyze how this environment has behaved during the years 1988, 1994 and 2006. The definition of that, there was the fact that during that period, which occurred in that area and more intensive space exploration to practice economic activities as well as a greater and more rapid expansion of urban area. This study was supported by the Remote Sensing, who have been shown today, as an efficient analysis of the environmental studies, through geoprocessing techniques. From the performed analysis it was found that occurred during the period of study (1988 to 2006), a huge change in the design of the estuary of Potengi river. The figures showed the area of vegetation was decreased 65.22%, the deforested area increased by 70.44% and the shrimp activity grow 452.07% and the urban area was increased in 52.65% along that period described. Considering the numbers shown there that the process of occupying space in that area requires attention, because it is an environment that represents a huge contribution to the ecological balance.

Alternative Transfers to the NEOs 99942 Apophis, 1994 WR12, and 2007 UW1 via Derived Trajectories from Periodic Orbits of Family G

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

Interplanetary natural routes via Moon and Lagrangiam points L1 and L2

Swing-by techniques are extensively used in interplanetary missions to minimize fuel consumption and to raise payloads of spaceships. The effectiveness of this type of maneuver has been proven since the beginning of space exploration. According to this premise, we have explored the existence of a natural and direct links between low Earth orbits and the lunar sphere of influence, to obtain low-energy interplanetary trajectories through swing-bys with the Moon and the Earth. The existence of these links are related to a family of retrograde periodic orbits around the Lagrangian equilibrium point L1 predicted for the circular, planar, restricted three-body Earth-Moon-particle problem. The trajectories in these links are sensitive to small disturbances. This enables them to be conveniently diverted reducing so the cost of the swing-by maneuver. These maneuvers allow us a gain in energy sufficient for the trajectories to escape from the Earth-Moon system and to stabilize in heliocentric orbits between the Earth and Venus or Earth and Mars. On the other hand, still within the Earth sphere of influence, and taking advantage of the sensitivity of the trajectories, is possible to design other swing-bys with the Earth or Moon. This allows the trajectories to have larger reach, until they can reach the orbit of other planets as Venus and Mars.(3σ)Broucke, R.A., Periodic Orbits in the Restricted Three-Body Problem with Earth-Moon Masses, JPL Technical Report 32-1168, 1968.

Using multiple abstraction levels to speedup an MPSoC virtual platform simulator

Virtual platforms are of paramount importance for design space exploration and their usage in early software development and verification is crucial. In particular, enabling accurate and fast simulation is specially useful, but such features are usually conflicting and tradeoffs have to be made. In this paper we describe how we integrated TLM communication mechanisms into a state-of-the-art, cycle-accurate, MPSoC simulation platform. More specifically, we show how we adapted ArchC fast functional instruction set simulators to the MPARM platform in order to achieve both fast simulation speed and accuracy. Our implementation led to a much faster hybrid platform, reaching speedups of up to 2.9 and 2.1x on average with negligible impact on power estimation accuracy (average 3.26% and 2.25% of standard deviation). © 2011 IEEE.

The aster mission: Exploring for the first time a triple system asteroid

2001 SN263 is a triple system asteroid. Although it was discovery in 2001, in 2008 astronomical observation carried out by Arecibo observatory revealed that it is actually a system with three bodies orbiting each other. The main central body is an irregular object with a diameter about 2.8 km, while the other two are small objects with less than 1 km across. This system presents an orbital eccentricity of 0.47, with perihelion of 1.04 and aphelion of 1.99, which means that it can be considered as a Near Earth Object. This interesting system was chosen as the target for the Aster mission - first Brazilian space exploration undertaking. A small spacecraft with 150 kg of total mass, 30 kg of payload with 110 W available for the instruments, is scheduled to be launched in 2015, and in 2018 it will approach and will be put in orbit of the triple system. This spacecraft will use electric propulsion and in its payload it will carry image camera, laser rangefinder, infrared spectrometer, mass spectrometer, and experiments to be performed in its way to the asteroid. This mission represents a great challenge for the Brazilian space program. It is being structured to allow the full engagement of the Brazilian universities and technological companies in all the necessary developments to be carried out. In this paper, we present some aspects of this mission, including the transfer trajectories to be used, and details of buss and payload subsystems that are being developed and will be used. Copyright ©2010 by the International Astronautical Federation. All rights reserved.