摘要:

AbstractThe problem or emulating multilevel structures on hypercube supercomputers is studied in the paper. The frequently used pyramid belongs to the class of multilevel structures. Several algorithms have been published in the literature for the emulation of pyramids by hypercubes. The paper extends the most important of these algorithms to make them applicable for multilevel structures. Results for the Connection Machine system CM‐2 with 16,384 processors are presented and comparative analysis of the new algorithms is carried out. It is shown that very often higher performance can be obtained for multilevel structures other than the pyrami

ISSN:1040-3108    

DOI:10.1002/cpe.4330060202

出版商:John Wiley&Sons, Ltd    

年代:1994

数据来源: WILEY

摘要:

AbstractIf problems involving unstructured meshes are to be solved efficiently on distributed‐memory parallel computers, the meshes must be partitioned and distributed across processors in a way that balances the computational load and minimizes communication. The recursive spectral bisection method (RSB) has been shown to be very effective for such partitioning problems compared to alternative methods, but RSB in its simplest form is expensive. Here a multilevel version of RSB is introduced that attains about an order‐of‐magnitude improvement in run time on typical exa

ISSN:1040-3108    

DOI:10.1002/cpe.4330060203

出版商:John Wiley&Sons, Ltd    

年代:1994

数据来源: WILEY

摘要:

AbstractThe paper presents a Conservative Time Window (CTW) algorithm for parallel simulation of discrete event systems. The physical system to be simulated is partitioned intondisjoint sub‐systems, each of which is represented by an object The CTW algorithm identifies a time window for each object, such that events occurring in each window are independent of events in other windows and thus they can be processed concurrently. The CTW algorithm was implemented on a shared memory multiprocessor, a Sequent Symmetry S81 with 16 processors. We measured performance of the CTW algorithm on two types of network topologies: feed‐forward networks and networks with feedback loops. We used three metrics to measure performance: speed‐up, average number of independent windows detected by the algorithm, and average number of events occurring in each window. We also investigated the impact of various event scheduling policies on performance. The results obtained show that the CTW algorithm produces good performance in many

ISSN:1040-3108    

DOI:10.1002/cpe.4330060204

出版商:John Wiley&Sons, Ltd    

年代:1994

数据来源: WILEY

摘要:

AbstractWe develop a balanced, parallel quicksort algorithm for a hypercube and compare it with a similar algorithm for a binary tree machine. The performance of the hypercube algorithm is measured on a Computing Surface.

ISSN:1040-3108    

DOI:10.1002/cpe.4330060205

出版商:John Wiley&Sons, Ltd    

年代:1994

数据来源: WILEY

ISSN:1040-3108    

DOI:10.1002/cpe.4330060201

出版商:John Wiley&Sons, Ltd    

年代:1994

数据来源: WILEY