Distributed Region-Based Memory Allocation and Synchronization

We present DRASync, a region-based allocator that implements a global address space abstraction for MPI programs with pointer-based data structures. The main features of DRASync are: (a) it amortizes communication among nodes to allow efficient parallel allocation in a global address space; (b) it takes advantage of bulk deallocation and good locality with pointer-based data structures; (c) it supports ownership semantics of regions by nodes akin to reader–writer locks, which makes for a high-level, intuitive synchronization tool in MPI programs, without sacrificing message-passing performance. We evaluate DRASync against a state-of-the-art distributed allocator and find that it produces comparable performance while offering a higher-level abstraction to programmers.

Dynamic Job Scheduling on the Grid Environment using the Great Deluge Algorithm

The utilization of the computational Grid processor network has become a common method for researchers and scientists without access to local processor clusters to avail of the benefits of parallel processing for compute-intensive applications. As a result, this demand requires effective and efficient dynamic allocation of available resources. Although static scheduling and allocation techniques have proved effective, the dynamic nature of the Grid requires innovative techniques for reacting to change and maintaining stability for users. The dynamic scheduling process requires quite powerful optimization techniques, which can themselves lack the performance required in reaction time for achieving an effective schedule solution. Often there is a trade-off between solution quality and speed in achieving a solution. This paper presents an extension of a technique used in optimization and scheduling which can provide the means of achieving this balance and improves on similar approaches currently published.

The Cost of Sustainability Capital and the Creation of Sustainable Value by Companies

We develop and apply a valuation methodology to calculate the cost of sustainability capital, and, eventually, sustainable value creation of companies. Sustainable development posits that decisions must take into account all forms of capital rather than just economic capital. We develop a methodology that allows calculation of the costs that are associated with the use of different forms of capital. Our methodology borrows the idea from financial economics that the return on capital has to cover the cost of capital. Capital costs are determined as opportunity costs, that is, the forgone returns that would have been created by alternative investments. We apply and extend the logic of opportunity costs to the valuation not only of economic capital but also of other forms of capital. This allows (a) integrated analysis of use of different forms of capital based on a value-based aggregation of different forms of capital, (b) determination of the opportunity cost of a bundle of different forms of capital used in a company, called cost of sustainability capital, (c) calculation of sustainability efficiency of companies, and (d) calculation of sustainable value creation, that is, the value above the cost of sustainability capital. By expanding the well-established logic of the valuation of economic capital in financial markets to cover other forms of capital, we provide a methodology that allows determination of the most efficient allocation of sustainability capital for sustainable value creation in companies. We demonstrate the practicability of the methodology by the valuation of the sustainability performance of British Petroleum (BP).

Efficient computation algorithm for calculating load contributions to line flows and losses

In a deregulated power system, it is usually required to determine the shares of each load and generation in line flows, to permit fair allocation of transmission costs between the interested parties. The paper presents a new method of determining the contributions of each load to line flows and losses. The method is based on power-flow topology and has the advantage of being the least computationally demanding of similar methods.

Computationally efficient sequential learning algorithms for direct link resource-allocating networks

Computionally efficient sequential learning algorithms are developed for direct-link resource-allocating networks (DRANs). These are achieved by decomposing existing recursive training algorithms on a layer by layer and neuron by neuron basis. This allows network weights to be updated in an efficient parallel manner and facilitates the implementation of minimal update extensions that yield a significant reduction in computation load per iteration compared to existing sequential learning methods employed in resource-allocation network (RAN) and minimal RAN (MRAN) approaches. The new algorithms, which also incorporate a pruning strategy to control network growth, are evaluated on three different system identification benchmark problems and shown to outperform existing methods both in terms of training error convergence and computational efficiency. (c) 2005 Elsevier B.V. All rights reserved.

Hicks meets Hotelling: the direction of technical change in capital-resource economies

We analyze a two-sector growth model with directed technical change where man-made capital and exhaustible resources are essential for production. The relative profitability of factor-specific innovations endogenously determines whether technical progress will be capital- or resource-augmenting. We show that any balanced growth equilibrium features purely resource-augmenting technical change. This result is compatible with alternative specifications of preferences and innovation technologies, as it hinges on the interplay between productive efficiency in the final sector, and the Hotelling rule characterizing the efficient depletion path for the exhaustible resource. Our result provides sound micro-foundations for the broad class of models of exogenous/endogenous growth where resource-augmenting progress is required to sustain consumption in the long run, contradicting the view that these models are conceptually biased in favor of sustainability.

Power-Capped DVFS and Thread Allocation with ANN Models on Modern NUMA Systems

Abstract—Power capping is an essential function for efficient power budgeting and cost management on modern server systems. Contemporary server processors operate under power caps by using dynamic voltage and frequency scaling (DVFS). However, these processors are often deployed in non-uniform memory
access (NUMA) architectures, where thread allocation between cores may significantly affect performance and power consumption. This paper proposes a method which maximizes performance under power caps on NUMA systems by dynamically optimizing two knobs: DVFS and thread allocation. The method selects the optimal combination of the two knobs with models based on artificial neural network (ANN) that captures the nonlinear effect of thread allocation on performance. We implement
the proposed method as a runtime system and evaluate it with twelve multithreaded benchmarks on a real AMD Opteron based NUMA system. The evaluation results show that our method outperforms a naive technique optimizing only DVFS by up to
67.1%, under a power cap.