Abstract-Sparse tensors appear in many large-scale applications with multidimensional and sparse data. While multidimensional sparse data often need to be processed on manycore processors, attempts to develop highly-optimized GPU-based implementations of sparse tensor operations are rare. The irregular computation patterns and sparsity structures as well as the large memory footprints of sparse tensor operations make such implementations challenging. We leverage the fact that sparse tensor operations share similar computation patterns to propose a unified tensor representation called F-COO. Combined with GPU-specific optimizations, F-COO provides highly-optimized implementations of sparse tensor computations on GPUs. The performance of the proposed unified approach is demonstrated for tensor-based kernels such as the Sparse Matricized TensorTimes-Khatri-Rao Product (SpMTTKRP) and the Sparse TensorTimes-Matrix Multiply (SpTTM) and is used in tensor decomposition algorithms. Compared to state-of-the-art work we improve the performance of SpTTM and SpMTTKRP up to 3.7 and 30.6 times respectively on NVIDIA Titan-X GPUs. We implement a CANDECOMP/PARAFAC (CP) decomposition and achieve up to 14.9 times speedup using the unified method over state-of-the-art libraries on NVIDIA Titan-X GPUs.

Grid is a distributed high performance computing paradigm that offers various types of resources (like computing, storage, communication) to resource-intensive user tasks. These tasks are scheduled to allocate available Grid resources efficiently to achieve high system throughput and to satisfy A. Y. Zomaya University of Sydney, Sydney, Australia user requirements. The task scheduling problem has become more complex with the ever increasing size of Grid systems. Even though selecting an efficient resource allocation strategy for a particular task helps in obtaining a desired level of service, researchers still face difficulties in choosing a suitable technique from a plethora of existing methods in literature. In this paper, we explore and discuss existing resource allocation mechanisms for resource allocation problems employed in Grid systems. The work comprehensively surveys Gird resource allocation mechanisms for different architectures (centralized, distributed, static or dynamic). The paper also compares these resource allocation mechanisms based on their common features such as time complexity, searching mechanism, allocation strategy, optimality, operational environment and objective function they adopt for solving computing-and data-intensive applications. The comprehensive analysis of cutting-edge research in the Grid domain presented in this work provides readers with an understanding of essential concepts of resource allocation mechanisms in Grid systems and helps them identify important and outstanding issues for further investigation. It also helps readers to choose the most appropriate mechanism for a given system/application.

A wide class of finite-element (FE) electromagnetic applications requires computing very large sparse matrix vector multiplications (SMVM). Due to the sparsity pattern and size of the matrices, solvers can run relatively slowly. The rapid evolution of graphic processing units (GPUs) in performance, architecture, and programmability make them very attractive platforms for accelerating computationally intensive kernels such as SMVM. This work presents a new algorithm to accelerate the performance of the SMVM kernel on graphic processing units.Index Terms-Computer architecture, graphic processing units (GPUs), parallel processing, sparse matrix vector multiplication (SMVM).

No abstract

Abstract-A study of the fundamental obstacles to accelerate the preconditioned conjugate gradient method on modern graphic processing units (GPUs) is presented and several techniques are proposed to enhance its performance over previous work. The performance increase observed when implementing one of our proposed enhancements over traditional approaches is reported.