Results 1 
5 of
5
Empirical Analysis of Overheads in Cluster Environments
 CONCURRENCY: PRACTICE AND EXPERIENCE
, 1995
"... In concurrent computing environments that are based on heterogeneous processing elements interconnected by generalpurpose networks, several classes of overheads contribute to lowered performance. In an attempt to gain a deeper insight into the exact nature of these overheads, and to develop stra ..."
Abstract

Cited by 24 (5 self)
 Add to MetaCart
(Show Context)
In concurrent computing environments that are based on heterogeneous processing elements interconnected by generalpurpose networks, several classes of overheads contribute to lowered performance. In an attempt to gain a deeper insight into the exact nature of these overheads, and to develop strategies to alleviate them, we have conducted empirical studies of selected applications representing different classes of concurrent programs. These analyses have identified load imbalance, the parallelism model adopted, communication delay and throughput, and system factors as the primary factors affecting performance in cluster environments. Based on the degree to which these factors affect specific classes of applications, we propose a combination of model selection criteria, partitioning strategies, and software system heuristics to reduce overheads and enhance performance in network based environments. We demonstrate that agenda parallelism and load balancing strategies contribu...
Fast Stable Merging And Sorting In Constant Extra Space
, 1990
"... In an earlier research paper [HL1], we presented a novel, yet straightforward lineartime algorithm for merging two sorted lists in a fixed amount of additional space. Constant of proportionality estimates and empirical testing reveal that this procedure is reasonably competitive with merge routines ..."
Abstract

Cited by 8 (0 self)
 Add to MetaCart
In an earlier research paper [HL1], we presented a novel, yet straightforward lineartime algorithm for merging two sorted lists in a fixed amount of additional space. Constant of proportionality estimates and empirical testing reveal that this procedure is reasonably competitive with merge routines free to squander unbounded additional memory, making it particularly attractive whenever space is a critical resource. In this paper, we devise a relatively simple strategy by which this efficient merge can be made stable, and extend our results in a nontrivial way to the problem of stable sorting by merging. We also derive upper bounds on our algorithms' constants of proportionality, suggesting that in some environments (most notably external file processing) their modest runtime premiums may be more than offset by the dramatic space savings achieved.
Parallel methods for Solving Fundamental File Rearrangement Problems
, 1990
"... We present parallel algorithms for the elementary binary set operations that, given an EREW PRAM with k processors, operate on two sorted lists of total length n in O(n=k + log n) time and O(k) extra space, and are thus timespace optimal for any value of k n=(log n). Our methods are stable, requir ..."
Abstract
 Add to MetaCart
We present parallel algorithms for the elementary binary set operations that, given an EREW PRAM with k processors, operate on two sorted lists of total length n in O(n=k + log n) time and O(k) extra space, and are thus timespace optimal for any value of k n=(log n). Our methods are stable, require no information other than a record's key, and do not modify records as they execute. ii Symbols Used O capital Greek omicron of "big oh" notation 6 capital Greek sigma for summations [ set union " set intersection 8 set exclusive or iii 1. Introduction The design and analysis of optimal parallel file rearrangement algorithms has long been a topic of widespread attention. The vast majority of the published literature has concentrated on the search for algorithms that are time optimal , that is, those that achieve optimal speedup (see, for example, [AS]). Unfortunately, space management issues have often taken a back seat in these efforts, leaving those who seek to implement optima...
Parallel Benchmarks and ComparisonBased Computing
, 1995
"... Nonnumeric algorithms have been largely ignored in parallel benchmarking suites. Prior studies have concentrated mainly on the computational speed of processors within very regular and structured numeric codes. In this paper, we survey the current state of nonnumeric benchmark algorithms and inves ..."
Abstract
 Add to MetaCart
(Show Context)
Nonnumeric algorithms have been largely ignored in parallel benchmarking suites. Prior studies have concentrated mainly on the computational speed of processors within very regular and structured numeric codes. In this paper, we survey the current state of nonnumeric benchmark algorithms and investigate the use of inplace merging as a suitable candidate for this role. Inplace merging enjoys several important advantages, including the scalability of efficient memory utilization, the generality of comparisonbased computing and the representativeness of nearrandom data access patterns. Experimental results over several families of parallel architectures are presented. A preliminary version of a portion of this paper was presented at the International Conference on Parallel Computing held in Gent, Belgium, in September, 1995. This research has been supported in part by the National Science Foundation under grant CDA9115428 and by the Office of Naval Research under contract N00014...