The WaveScalar is the first dataflow architecture that can efficiently provide the sequential memory semantics required by imperative languages. This work presents a speculative memory disambiguation mechanism for thi...
详细信息
The WaveScalar is the first dataflow architecture that can efficiently provide the sequential memory semantics required by imperative languages. This work presents a speculative memory disambiguation mechanism for this architecture, the transaction WaveCache. Our mechanism maintains the execution order of memory operations within blocks of code, called waves, but adds the ability to speculatively execute, out-of-order, operations from different waves. This mechanism is inspired by progress in supporting transactional memories. Waves are considered as atomic regions and executed as nested transactions. Wave that have finished the execution of all their memory operations are committed, as soon as the previous waves are also committed. If a hazard is detected in a speculative wave, all the following waves (children) are aborted and re-executed. We evaluated the transactional WaveCache on a set of benchmarks from Spec 2000, Mediabench and Mibench (telecomm). Speedups ranging from 1.31 to 2.24 (related to the original WaveScalar) where observed when the benchmark doesn't perform lots of emulated function calls or access memory very often. Low speedups of 1.1 to slowdowns of 0.96 were observed when the opposite happens or when the memory concurrency was high.
We consider a scenario in which users share an access point and are mainly interested in VoIP applications. Each user is allowed to adapt to varying network conditions by choosing the transmission rate at which VoIP t...
详细信息
We consider a scenario in which users share an access point and are mainly interested in VoIP applications. Each user is allowed to adapt to varying network conditions by choosing the transmission rate at which VoIP traffic is received. We denote this adaptation process by end-user congestion control, our object of study. The two questions that we ask are: (1) what are the performance consequences of letting the users to freely choose their rates? and (2) how to explain the adaptation process of the users? We set a controlled lab experiment having students as subject to answer the first question, and we extend an evolutionary game-theoretic model to address the second. Our partial answers are the following: (1) free users with local information can reach an equilibrium which is close to optimal from the system perspective. However, the equilibrium can be unfair; (2) the adaptation of the users can be explained using a game theoretic model. We propose a methodology to parameterize the latter, which involves active network measurements, simulations and an artificial neural network to estimate the QoS perceived by the users in each of the states of the model.
In this paper we present the architecture for the Personal Autonomic Desktop Manager, a self managing application designed to act on behalf of the user in several aspects: protection, healing, optimization and configu...
详细信息
In this paper we present the architecture for the Personal Autonomic Desktop Manager, a self managing application designed to act on behalf of the user in several aspects: protection, healing, optimization and configuration. The overall goal of this research is to improve the correlation of the autonomic self{sup}* properties and doing so also enhance the overall self-management capacity of the desktop (autonomicity). We introduce the Circulatory Computing (CC) model, a self-managing system initiative based on the biological metaphor of the cardiovascular system, and use its concepts in the design and implementation of the architecture.
One of the most known applications of Discrete Optimization is on scheduling. In contrast, one of the most known applications of Continuous Nonlinear Optimization is on the control of dynamic systems. In this paper, w...
详细信息
ISBN:
(纸本)9789955282839
One of the most known applications of Discrete Optimization is on scheduling. In contrast, one of the most known applications of Continuous Nonlinear Optimization is on the control of dynamic systems. In this paper, we combine both views, solving scheduling problems as dynamic systems, modeled as discrete-time nonlinear optimal control problems with state and control continuous variables subjected to upper and lower bounds. The proposed formulation has the following advantages over discrete (mixedinteger) models: a smaller number of variables is employed, and no 0-1 variable is needed. Therefore, the scheduling problem can be solved as a standard continuous nonlinear program. Complementarity constraints are used to represent scheduling decisions, defining a nonconvex problem, which can be solved with Global Optimization (GO) and Nonlinear programming (NLP) methods. Applications with a continuous process background are discussed, such as the ones from petroleum and water & wastewater industries, because they pose challenging issues, with a combination of nonlinear and combinatorial aspects. One example we discuss in detail is the crude oil scheduling in ports, with tanks, pipelines, jetties, and tanker vessels and blending operations. The recent literature on this problem is rich in mixedinteger linear programming (MILP) models, therefore we developed a procedure to reformulate certain mixed-integer constraints as complementarity constraints, discarding the associated binary variables. The resulting NLP model is equivalent to the original MILP, in a sense that a feasible point in the NLP is also a feasible point in the MILP. A number of numerical cases are discussed to illustrate the validity of this approach. Despite obtaining good results with the NLP approach, we acknowledge that the MILP has the desirable feature of having only global optima, whereas the NLP is non-convex. Therefore, we present an hybrid NLP-MILP scheme that uses the NLP to generate new MILP inte
Voltage spikes are ubiquitous in biological nervous systems. How spikes can be used to encode signals, facilitate communication, and implement important computations is an important question of contemporary neuroscien...
详细信息
Voltage spikes are ubiquitous in biological nervous systems. How spikes can be used to encode signals, facilitate communication, and implement important computations is an important question of contemporary neuroscience. Acoustic processing tasks provide a rich range of applications for this encoding scheme. As a summary of the Ph.D. research of the first author, we present two analog VLSI spike-based example systems that process acoustic information using spikes: a model of the neural signal processing involved in bat echolocation, and a low-power, time-domain acoustic periodicity detector.
To support our ongoing work in modeling bat echolocation, an artificial bat head was designed and fabricated using a 3D printer, an ultrasonic cochlea-like filter bank with 16 channels was designed with moderate quali...
详细信息
To support our ongoing work in modeling bat echolocation, an artificial bat head was designed and fabricated using a 3D printer, an ultrasonic cochlea-like filter bank with 16 channels was designed with moderate quality (Q) factor, and 128 spiking neurons convert these signals to spike trains. A two-dimensional address-event arbiter is used to transmit these spikes off of the chip. We demonstrate that the population of spiking neurons can be decoded to estimate azimuth and elevation of ultrasonic chirps. This chip was fabricated in a commercially-available 0.5(mu)m CMOS process and consumes approximately 36(mu)W.
The third international conference on Human-Robot Interaction (HRI-2008) was held in Amsterdam, The Netherlands, March 12-15, 2008. The theme of HRI-2008, "living with robots," highlights the importance of t...
详细信息
Recommendation accuracy is especially important in mobile e-commerce environments due to the limited screen size of mobile devices and relatively expensive connection costs. Mobile content tends to be fashionable and ...
详细信息
ISBN:
(纸本)1601320639
Recommendation accuracy is especially important in mobile e-commerce environments due to the limited screen size of mobile devices and relatively expensive connection costs. Mobile content tends to be fashionable and are geared for young users. This paper presents a novel method of building a more accurate recommender system for mobile content in a mobile ecommerce environment. The method is based on collaborative filtering, and models content diffusion and user preference transition and incorporates them in constructing pseudo ratings from implicit feedback data. In a variety of experiments, recommender systems based on the method showed significantly better recommendation accuracy than a pure collaborative filtering-based recommender system.
Emerging 64 bitOSpsilas supply a huge amount of memory address space that is essential for new applications using very large data. It is expected that the memory in connected nodes can be used to store swapped pages e...
详细信息
Emerging 64 bitOSpsilas supply a huge amount of memory address space that is essential for new applications using very large data. It is expected that the memory in connected nodes can be used to store swapped pages efficiently, especially in a dedicated cluster which has a high-speed network such as 10 GbE and Infiniband. In this paper, we propose the distributed large memory system (DLM), which provides very large virtual memory by using remote memory distributed over the nodes in a cluster. The performance of DLM programs using remote memory is compared to ordinary programs using local memory. The results of STREAM, NPB and Himeno benchmarks show that the DLM achieves better performance than other remote paging schemes using a block swap device to access remote memory. In addition to performance, DLM offers the advantages of easy availability and high portability, because it is a user-level software without the need for special hardware. To obtain high performance, the DLM can tune its parameters independently from kernel swap parameters. We also found that DLMpsilas independence of kernel swapping provides more stable behavior.
暂无评论