Withthe rapid progress of high-performance cluster applications, data transfer between clusters in distant locations becomes more important. But, it is difficult to transfer data using parallel TCP streams on long di...
详细信息
the developers of distributed algorithms are faced with two opposing forces. One is to design generic algorithms that are reusable in a large number of applications. Efficiency considerations, on the other hand, force...
详细信息
ISBN:
(纸本)9780769534725
the developers of distributed algorithms are faced with two opposing forces. One is to design generic algorithms that are reusable in a large number of applications. Efficiency considerations, on the other hand, force the algorithms to be Customized to specific operational contexts. this problem is often attacked by simply re-implementing all or large portions of an algorithm. this paper proposes InDiGO, ail infrastructure which allows design of generic hut customizable algorithms and provides tools to customize such algorithms for specific applications. InDiGO provides the following capabilities: (a) Tools to generate intermediate representations of an application which can be leveraged for analysis, (b) Mechanisms to allow developers to design customizable algorithms by exposing design knowledge in terms of configurable options, and (c) Ail optimization engine to analyze an application to derive the information necessary optimize the algorithms. We perform three types of optimizations: static, dynamic and physical topology-based optimization. We present experimental results to demonstrate the advantages of our infrastructure.
this article describes a system for run-time prediction of applications in heterogeneous environments. To exploit the power of computational grids, scheduling systems need profound information about the job to be exec...
详细信息
ISBN:
(纸本)9780769534725
this article describes a system for run-time prediction of applications in heterogeneous environments. To exploit the power of computational grids, scheduling systems need profound information about the job to be executed. the run-time of a job is - beside others - not only dependent of its kind and complexity but also of the adequacy and load of the remote host where it will be executed. Accounting and billing are additional aspects that have to be considered when creating a schedule. Currently predictions are achieved by using descriptive models of the applications or by applying statistical methods to former jobs mostly neglecting the behaviour of users. Motivated by this, we propose a method that is not only based on the characteristics of a job but also takes the behaviour of single users and groups of similar users respectively into account. the basic idea of our approach is to cluster users, hosts and jobs and apply multiple methods in order to detect similarities and create forecasts. this is achieved by tagging jobs with attributes and by deriving predictions for similar attributed jobs whereas the recent behaviour of a user determines which predictions tire finally taken.
this paper presents a distributed file-system for the present day medium-sized network. Existing servers and workstations pool their unused storage resources to form a communal share. Erasure codes provide fault toler...
详细信息
While recently the focus of architects and programmers has been on multi core, the alternative of processor node plus array oriented accelerator has some significant advantages especially in Compute intensive static a...
详细信息
ISBN:
(纸本)9780769534725
While recently the focus of architects and programmers has been on multi core, the alternative of processor node plus array oriented accelerator has some significant advantages especially in Compute intensive static applications. We propose an acceleration methodology based on FPGA arrays (but, in principle it could be GPU or Cell based). the methodology, uses a comprehensive application analysis supported by high performance FPGA hardware. the analysis provides a dataflow graph of the application which is replicated in SIMD for multiple data strips (until limited by the pin bandwidth, then pipelined (MISD) until circuit limited. An oil exploration application shows the possibility of speedup of over 300x over an Intel Xeon.
this paper concerns speculative parallelization as a method of improving computations efficiency and also as a method of reducing the problem solving time with reference to its sequential version. Speculative parallel...
详细信息
ISBN:
(纸本)9780769534725
this paper concerns speculative parallelization as a method of improving computations efficiency and also as a method of reducing the problem solving time with reference to its sequential version. Speculative parallelization is proposed or a particular class of problems, described as recursive functions taking values from finite sets. It refers to speculative execution of consecutive iteration steps, each of which, except the first one, depends on the preceding iteration step yet before it ends. Assuming that in the sequential version one iteration is performed in one linear execution time step (hereinafter referred to as computational step), then the aim of the speculative parallelization is the reduction of the total number of computational steps and thus execution of more than one iteration in one time step. the essence of the problem is that we assume some mapping schemes of arguments into the set of possible values of the function in speculative computing, i.e. there exists precise information about the possible values that the function can take for particular arguments. this paper presents simulation results for the chosen mapping schemes, illustrating how the number of steps, required to compute the value of the function for the given argument, depends on the structure of the mapping scheme and the number of used parallelthreads.
In this paper we consider the frameworks of Process Algebra and I/O Automata and we apply both towards the verification of a distributed leader-election protocol. Based on the two experiences we evaluate the approache...
详细信息
distributed storage systems apply erasure-tolerant codes to guarantee reliable access to data despite failures of storage resources. While many codes can be mapped to XOR operations and efficiently implemented on comm...
详细信息
Bundle-Exchange-Compute (BEC) is a new virtual shared memory parallel programming environment for distributed-memory machines. Different from and complementary to other Global Address Space (GAS) programming model res...
详细信息
ISBN:
(纸本)9780769534725
Bundle-Exchange-Compute (BEC) is a new virtual shared memory parallel programming environment for distributed-memory machines. Different from and complementary to other Global Address Space (GAS) programming model research efforts, BEC has built-in efficient support for unstructured applications that inherently require high-volume random fine-grained communication, such as parallel graph algorithms, sparse-matrices, and large-scale physics simulations. In BEC, the global view of shared data structures enables ease of algorithm design and programming;and for good application performance, fine-grained (random) accesses to shared data are automatically and dynamically bundled together for coarse-grained message-passing. BEC frees the users from explicit management of data distribution, locality, and communication. therefore, BEC is much easier to program than MPI, while achieving comparable application performance. this paper presents some initial BEC applications, which show that simple BEC programs can match very complex and highly optimized MPI codes.
暂无评论