Titlebook: High Performance Computing - HiPC‘99; 6th International Co Prith Banerjee,Viktor K. Prasanna,Bhabani P. Sinha Conference proceedings 1999 S

责问

Harnessing Windows NT for High Performance Computingsing has also gained widespread acceptance. The ever increasing speed of the microprocessor and the availability of economically viable high speed interconnection networks combine to offer a powerful and inexpensive solution to parallel processing using a cluster of workstations. Windows NT offers a

Cacophonous

隐士

Modeling Cone-Beam Tomographic Reconstruction Using LogSMP: An Extended LogP Model for Clusters of Sate interesting objects. The analysis of its parallel implementation on widely available clusters of SMPs requires an extension of the original LogP model to account for the various communication channels, called LogSMP. The LogSMP model is used in analyzing this algorithm, which predicts speedup on

Graduated

驾驶

爱国者

Memory-Optimal Evaluation of Expression Trees Involving Large Objectsften, the tree node objects are very large that only a subset of them can fit in memory at a time. This paper addresses the problem of finding an evaluation order of nodes in a given expression tree that uses the least memory. We develop an efficient algorithm that finds an optimal evaluation order

IDEAS

Resource Usage Modelling for Software Pipeliningnique to reduce the space overhead in automaton based software pipelining methods. This technique exploits the symmetry in the states of the automaton to produce the Group Automaton (GA) with reduced number of states. Our second technique is the Conjugate Offset method which eliminates the “symmetri

URN

结构

Fast Slicing of Concurrent Programsre of the intermediate graph representation of concurrent programs and this leads us to introduce the notion of Concurrent Control Flow Graph (CCFG). We introduce a parallel algorithm to compute static slices of concurrent programs using CCFG.

形状

Performance Evaluation of a Load Sharing System on a Cluster of Workstationsists of successive manipulations of algorithms parameters until reaching satisfied approximations. Once the suitable parameters have been tuned, which configure the specific model, it is easy to evaluate the performance of the load allocating algorithms for various applications and hardware configurations.